Light-driven bioprocesses

3.1 Indirect photoactivation of redox enzymes

In photobiocatalysis, several concepts have been reported that demonstrate the feasibility of reducing NAD(P)⁺ by exploiting the reduction potential of photosensitizers in their excited state [32]. Analogous to oxygenic photosynthesis, light is converted into chemical energy stored in NAD(P)H. Unlike the oxidation of NAD(P)H to NAD(P)⁺, the back-reaction to regenerate reduced nicotinamide has to be catalyzed regioselectively, as the direct reduction of NAD(P)⁺ can yield inactive isomers or undesired dimerization [33, 34]. Other studies suggest the use of additional organometallic compounds as electron mediators, which facilitate the formation of the active 1,4-NAD(P)H, subsequently consumed in enzymatic reactions [35, 36]. Therein, Rh^III and Ir^III complexes turned out to be highly potent in performing the mediation of electrons between an electrode and the cofactor, while a hydride ion (H⁻) serves as a reducing equivalent in the selective reduction of NAD(P)⁺ [36]. Based on these observations, indirect photochemical regeneration strategies of reduced NAD(P) have been developed (Scheme 2A), wherein the photochemical nicotinamide cofactor recycling using electron mediators is for instance coupled to the activity of cytochrome P450s [37] or dehydrogenases [28].

Scheme 2:

General photobiocatalytic concept for indirect activation of redox enzymes. A) General scheme of the indirect photoactivation redox enzymes. B) A thermophilic ene-reductase (TOYE) catalyzes the enzymatic reduction of an α,β-unsaturated substrate. The reaction is driven by photoexcited electrons deriving from the sacrificial electron donor triethanolamine (TEA). [Ru/bpz)₃]²⁺ is used as photosensitizer and methyl viologen (MV²⁺) as mediator [39]. C) Light-driven activation of cytochrome P450 BM3 catalyzing the O-dealkylation of 7-ethoxycoumarin [37]. Eosin Y is used as photosensitizer transferring electrons from the sacrificial electron donor TEOA to an organometallic mediator for regenerating NADPH. Adapted with permission from Ref. [16].

Early studies on photochemical NAD(P)H regeneration systems rely on the use of Ru[(bpy)₃]²⁺ or ZnTMPyP4 as photosensitizers in combination with methyl viologen (MV) as a primary electron acceptor [38]. In the respective photoinduced charge separation at the photosensitizer, electrons are initially transferred to MV²⁺, using either EDTA, 2-mercaptoethanol, triethylamine (TEA) or TEOA as sacrificial electron donors. The generated MV^+● radical acts as the electron donor for the subsequent enzymatic reduction of NAD(P)⁺ catalyzed by an NAD(P)⁺-dependent dehydrogenase (Scheme 2B) [39]. By coupling the light-driven cofactor regeneration system to a second enzymatic conversion, CO₂ conversion into keto acids [40], reduction of ketones [41], and reduction and amination of keto acids was demonstrated [38, 42]. However, the need for an additional biocatalyst to catalyze the reduction of NAD(P)⁺ does not offer a significant advantage over conventional enzymatic cofactor regeneration systems [31].

In searching for more safe and straightforward solutions for the photochemical regeneration of NAD(P)H, reaction systems implementing organic dyes [43], flavins [44], quantum dots [45], graphene-based photocatalysts [46] and porphyrins [47] as photosensitizers in combination with [Cp*Rh(bpy)H₂O]²⁺ as electron mediator were developed. [Cp*Rh(bpy)H₂O]²⁺ represents a versatile organometallic mediator exhibiting high activities and stability for the highly selective for generating the active 1,4-cofactor variants [15, 19, 44]. For example, Lee et al. used a light-driven approach to catalyze the O-dealkylation of 7-ethoxycoumarin with cytochrome P450 BM3 (Scheme 2C) [37]. The xanthene dye eosin Y acts as a photosensitizer to transfer photoexcited electrons from TEOA to an organometallic mediator molecule for NADPH regeneration. Lee et al. also proposed the light-driven regeneration of NADH by using the sacrificial electron donor TEOA and eosin Y as photosensitizer [28]. Eosin Y, an organic dye previously used for cell staining in histological applications, revealed its potential as a photosensitizer in various photochemical applications [48]. In a photobiocatalytic setup for the conversion of α-ketoglutarate to L-glutamate catalyzed by L-glutamate dehydrogenase, [Cp*Rh(bpy)H₂O]²⁺ was used as an electron mediator between photoexcited eosin Y and NAD⁺ yielding NADH [28]. Thereby, a turnover number of about 25 NAD(P)⁺ h⁻¹ was estimated for the photosensitizer-mediator driven cofactor regeneration system, yielding about 90 % substrate conversion. Driven by these findings, the scope of potential photosensitizers was further explored by screening different xanthene dyes [49]. As a result, rose bengal, fluorescein, erythrosine B and phloxine B show similar or even better performance in the light-driven cofactor supply for the GDH compared to the previously reported eosin Y.

Other examples report the use of graphene-based photosensitizers harboring covalently bound porphyrin derivates [46], or difluoro-boraindacene (BODIPY) [50], as chromophores that can donate electrons to [Cp*Rh(bpy)H₂O]²⁺ upon visible light exposure. Analogous to previously described examples, [Cp*Rh(bpy)H₂O]²⁺ catalyzes the reduction of NAD⁺ to NADH, the final step in the photocatalytic cofactor recycling mechanism. Also quantum dots have been used as photosensitizers in photochemical NADH cofactor regeneration systems with [Cp*Rh(bpy)H₂O]²⁺ [51].

The previous examples emphasize the feasibility of light-driven NAD(P)H regeneration combined with the enzymatic reduction of organic compounds. Similarly, light-induced charge separation and electron transfer can be utilized for the in situ generation of oxidized NAD(P)⁺. In this case, NAD(P)H generated during enzyme-catalyzed substrate oxidation serves as an initial electron donor quenching the oxidized photosensitizer. Consequently, an electron acceptor, often O₂, represents the final component in the light-driven electron transfer chain. Gargiulo et al. successfully coupled the photochemical generation of NAD(P)⁺ to the oxidation of alcohols into ketones and lactones catalyzed by alcohol dehydrogenase [52]. In this case, flavins were applied as photosensitizers for the light-driven reoxidation of NAD(P)H to NAD(P)⁺. Since oxygen serves as the terminal electron acceptor for the regeneration of the reduced flavin, the strong oxidant H₂O₂ is formed as a byproduct. While Gargiulo et al. used catalase to avoid possible side effects caused by H₂O₂, the use of dithionite as a photocatalytic component in the regeneration of NADP⁺ yields in the final formation of H₂O as proposed by Rickus et al. [53]. The generation of NADP⁺ was determined by the enzymatic conversion of isocitrate to α-ketoglutarate. The reaction catalyzed by an NADP⁺-specific isocitrate dehydrogenase revealed turnover frequencies of about 35 h⁻¹ by providing the photosensitizer in solution.

3.2 Direct photoactivation of redox enzymes

Reaction systems with the purpose of direct photoactivation of oxidoreductases have the following aims: activating cofactor-independent enzymes, circumventing the need for cofactors and decreasing the complexity of multicomponent reaction systems (Scheme 3A). Therefore, the focus is on the redox-active prosthetic groups, including flavins, heme moieties, and non-heme bond metal ions within the polypeptide framework of oxidoreductases. In nature, these enzymes receive electrons directly from reduced cofactors or indirectly via redox partners. These mechanisms provide a site for injecting photoexcited electrons resulting in light-driven enzyme activation.

Scheme 3:

Photobiocatalytic concepts for the direct activation of redox enzymes. A) General scheme for the direct photoactivation of redox enzymes via electron transfer from a sacrificial electron donor. B) In vitro direct photoactivation of an OYE variant from Thermus scotoductus (TsOYE) catalyzing the asymmetric reduction of C=C bonds [54]. Rose bengal as photosensitizer transfers photoexcited electrons to the prosthetic FMN group of TsOYE. C) The combination of a photosensitizer (e.g. 5(6)-carboxyeosin) with sacrificial electron donors (e.g. MES buffer) fuels Rieske oxygenase (RO)-catalyzed hydroxylations in whole cells [27]. In this example, photoexcited electrons are transferred to the active site (non-heme iron center) of the RO [27]. Adapted with permission from Ref. [16].

In this context, members of the ene-reductase (ERED) enzyme family have been studied for regeneration by chromophores that are being utilized as photosensitizers, which, upon light excitation, can transfer electrons from a sacrificial electron donor such as TEOA to the oxidized FMN within the active site of the EREDs (Scheme 3B) [54, 55]. Instead of indirectly regenerating the ERED via regeneration of reduced nicotinamide cofactors, reducing equivalents from simple sacrificial electron donors such as TEOA, EDTA, formate, or phosphite via were provided via photocatalytic oxidation. The photoenzymatic conversion of, for instance, 2-methylcyclohexenone by TsOYE in combination with rose bengal (RB) resulted in enantiopure (R)-2-methylcyclohexanone (ee >99 %) with a yield of up to 53 % and total turnover numbers of 235 for TsOYE (Scheme 3B) [56]. The substitution of halogen atoms in xanthene dyes significantly affected the turnover frequency.

Also heme-containing enzymes such as cytochrome P540s (CYPs) have been utilized for direct photoactivations. For instance, Park et al. demonstrated the light-induced electron transfer from TEOA to the heme-prosthetic group of various CYPs [57]. Emphasized as a cofactor-independent reaction system, the hydroxylation of various drugs and steroid molecules by using resting Escherichia coli cells containing overexpressed CYP variants and eosin Y as photosensitizer has been shown. In reactions performed for over 20 h under visible light irradiation, total turnover numbers of up to 180 of various CYPs were achieved. Analogously, a study conducted by Özgen et al. indicates the versatility of the described whole-cell approach using dye derivates as organic photosensitizers in combination with EDTA, MES or MOPS buffer as sacrificial electron donors (Scheme 3C) [27]. The asymmetric hydroxylation of olefins catalyzed by four Rieske non-heme iron-dependent oxygenase (RO) variants has been performed to confirm the applicability of the proposed reaction system to further multi-component oxygenases. By that, the costly supply for reduced NAD(P)H cofactors and limitations arising from the assumed instability of ROs in cell-free systems were circumvented.

3.3 Light-driven cofactor supply using photoautotrophs

The application of whole-cells as biocatalysts offers unique advantages and has been widely used in the biosynthesis of high-value fine and bulk chemicals, as well as active pharmaceutical ingredients [58, 59]. The continued identification of new microorganisms as well as recent advances in synthetic biology and metabolic engineering, together with the rapid development of molecular genetic tools, have led to a renaissance of whole-cell biocatalysis.

Besides upregulating the production of endogenous metabolites in industrial-scale fermentation processes [60], recombinant whole cells provide an interesting platform for biotransformations catalyzed by heterologous enzymes. By exploiting the cell’s metabolism, one-pot multi-step conversions of simple substrates into value-added chemicals without additional cofactor supply systems represent an overall goal in industrial biocatalysis [61]. However, the use of heterotrophic microorganisms such as Corynebacterium glutamicum in the industrial production of amino acids, or recombinant E. coli as a model organism for establishing e.g. enzyme cascades, is dependent on the sufficient supply of sugars (e.g. glucose) as carbon fuel for their metabolism. From an environmental perspective, glucose represents a renewable carbon source obtained from biomass and is thus considered a green alternative to petrochemical raw materials for producing fuels, bulk or fine chemicals. From a socio-economical perspective, however, glucose production is mainly based on the enzymatic hydrolysis of edible biomass, which is questioned as it overlaps with the global feedstock and food supply [62]. To counteract this debate, the development of alternative routes to increase the accessibility of fermentable carbon sources from non-edible biomass such as lignocellulose-rich materials has gained increased attention [63].

Unlike heterotrophs as whole-cell platforms for bioconversions, photoautotrophs can utilize CO₂ as a sole carbon source to fuel their metabolism. The energy needed to assimilate CO₂ is derived from natural light during photosynthesis [36]. The light-induced oxidation of water (referred to as water-splitting) yields hydrogen protons (H⁺) and electrons, and the light-generated electrons are directly used to reduce NADP⁺ to NADPH. This makes photosynthetic organisms a particularly interesting chassis for coupling heterologously expressed oxidoreductases to the rich NADPH pool, thereby driving whole-cell light-driven biotransformations [64]. Recently, cyanobacteria have emerged as highly interesting whole-cell biocatalysts in such light-driven redox reactions [64], and have been applied for instance in the asymmetric reduction of ketones [65] or aldehydes [66]. The similarity of cyanobacteria to other Gram-negative bacteria such as E. coli enables the generation of recombinant variants by applying available genetic engineering techniques [67]. By that, biotransformations can be conducted that go beyond the endogeneous reaction scope of cyanobacteria [68]. Considering the enhanced atom economy enabling water as a sacrificial electron donor to fuel enzymatic redox reactions, the use of cyanobacteria appears to have a significant advantage over recombinant heterotrophs.

Following these considerations, Köninger et al. coupled the photosynthetic apparatus of recombinant cyanobacteria to the activity of oxidoreductases (Figure 3A) [69]. Thereby, the asymmetric reduction of activated C=C bonds catalyzed by the NADPH-dependent ERED YqjM from Bacillus subtilis expressed in recombinant Synechocystis sp. PCC 6803 cells was successfully shown. On a semi-preparative scale using 100 mg substrate, an isolated product yield of up to 80 % of enantiopure (R)-2-methylsuccinimide was achieved, which is comparable to the efficiency of typical whole-cell biotransformations in E. coli [39]. Büchsenschütz et al. reported that recombinant Synechocystis cells can also be applied for the asymmetric reduction of cyclic amines [70]. Thereby, reduced NADPH cofactors, directly or indirectly provided by photosynthesis, can fuel the catalytic activity of imine reductases (IREDs) by supplying reducing equivalents. By optimizing reaction conditions, full conversions of prochiral substrates could be achieved, yielding optically pure secondary amines.

Figure 3:

The cyanobacterium Synechocystis sp. PCC 6803 is used in whole-cell biotransformations for A) the C=C double bond reduction catalyzed by the ene reductase YqjM from Bacillus subtilis [69, 74]. B) The hydroxylation of nonanoic acid methyl ester catalyzed by the monooxygenase AlkB from Pseudomonas putida Gpo1 [71, 72]. Adapted with permission from Refs. [69, 71] .

Hoschek et al. coupled the formation of O₂, driven by the initial water-splitting reaction of oxygenic photosynthesis, to enzyme-catalyzed oxyfunctionalization reactions (Figure 3B) [71]. It is assumed that the light-driven in situ generations of O₂ has the potential to counteract mass-transfer limitations between the gas and liquid phase, a crucial factor in the upscaling of O₂-dependent processes. In a corresponding study, engineered Synechocystis sp. PCC 6803 was used as a host organism for the heterologous expression of an NADH-dependent alkane monooxygenase (AlkB) from Pseudomonas putida GPo1 [72]. As revealed in whole-cell reactions performed under anaerobic conditions, the hydroxylation of nonanoic acid methyl ester into the corresponding ω-hydroxylated product strictly depends on the light-driven water-oxidation yielding a sufficient amount of intracellular O₂. Driven by these findings, the same research group evaluated the scalability of hydroxylation reactions without requiring external aeration [73]. The in vivo conversion of cyclohexane into the corresponding mono-hydroxylated alcohol catalyzed by a CYP from Acidovorax sp. in a two-liquid phase setup resulted in 2.6 g of cyclohexanol after 52 h.

3.4 Photocatalytic generation of hydrogen peroxide in situ

While the obtained turnover numbers for photobiocatalytic cofactor regeneration (e.g. NADPH, FAD) are in many cases too low for synthetic applications, the developed examples of combining a photosensitizer with a sacrificial electron donor can also be used for the in situ generation of H₂O₂. This is particularly interesting as many oxidoreductases such as CYPs and ROs rely on a complex electron-transfer chain and are cofactor-dependent, limiting their application in organic synthesis. In contrast, peroxygenases use H₂O₂ as oxidant, however they usually show poor robustness against high H₂O₂ concentrations. As such, the concentration of H₂O₂ to be carefully controlled to avoid rapid deactivation of the enzymes while maintaining high reactivity. In this context, several photocatalytic strategies for in situ H₂O₂ generation have been developed. For instance, Zhang et al. used gold–titanium dioxide (Au–TiO₂) to generate H₂O₂ through the methanol-driven reductive activation of ambient O₂ to drive an unspecific peroxygenase from Agrocybe aegerita (rAaeUPO) [75]. Using this approach, the stereoselective hydroxylation of ethylbenzene to (R)-1-phenylethanol was achieved with high enantioselectivity (>98 % ee) and high turnover numbers for the biocatalyst (>71,000). This reaction system was further explored by using water instead of methanol as electron donor to produce H₂O₂ in situ to drive rAaeUPO [76]. By using this system, the stereoselective hydroxylation of ethylbenzene to (R)-1-phenylethanol has also been achieved, resulting in 110 mg product with an ee of >97 %. This system was later on further optimized.

In addition to using TiO₂-based photocatalysts or even flavin [77] for the in situ H₂O₂ generation, further photocatalysts have been explored for the same purpose and enzyme class. In this regard, a water-soluble sodium anthraquinone sulfonate (SAS) was used to generate H₂O₂ to drive a vanadium-dependent chloroperoxidase from Curvularia inaequalis (CiV-CPO)-catalyzed halogenation reaction [78]. With this system, a yield of 91 % and turnover number of 318,000 was achieved. Interestingly, this approach eliminates the diffusion limitations of the heterogeneous photocatalyst and protects the enzyme from long-lived radicals. Other examples using nitrogen-doped carbon nanodots [79, 80] also highlight that these photocatalysts are promising to drive peroxygenase-catalyzed hydroxylation reactions. In particular, it was shown that the spatial separation of the photocatalyst from the enzyme avoids the inactivation of the enzyme, resulting in promising turnover numbers of the biocatalyst of more than 60,000 [79]. In addition, these nitrogen-doped carbon nanodots can also be applied in neat reaction media. For instance, it was shown that the immobilization of rAaeUPO increased enzyme stability and thus facilitated the reaction in neat substrate such as cyclohexane [80].

4.1 Natural photoenzymes

Yet, the scope of photoenzymes discovered in nature is very limited and the investigation of their biocatalytic applicability is often restricted to their natural substrate and reaction scope. One of the first natural photoenzymes was discovered in the 1950s by the physicist Claud S. Rupert [81]. This enzyme, a so-called photolyase, is a flavoprotein containing two light-harvesting cofactors, that is responsible for repairing UV light induced DNA damage (Scheme 4A) [82]. Although photolyases can catalyze DNA photoreactivation with considerable high efficiency, their implementation in biocatalytic applications has not yet been described in the literature [83].

Another class of natural photoenzymes are the light-dependent protochlorophyllide reductases (LPORs, Scheme 4B). These enzymes are involved in the biosynthesis of chlorophyll in both oxygenic and anoxygenic phototrophs, in which they catalyze the reduction of protochlorophyllide (pchlide) to chlorophyllide (chlide) [84]. Yet, the biocatalytic application of LPORs is limited, although it could be shown that in vitro they catalyze the conversion of various pchlide derivatives to the corresponding reduced products [85]. In addition, several LPOR homologs were successfully identified from different origins and biochemically characterized, and their activity was studied for the conversion of pchlide to chlide under different conditions and furthermore analyzed for their cofactor flexibility [86].

In 2017, Beisson et al. identified another natural photoenzyme from the microalgae Chlorella variabilis [87], and ever since its discovery, it is intensively studied for its biocatalytic potential. This enzyme is a fatty acid decarboxylase (FAP) catalyzing a light-mediated decarboxylation of fatty acids via a flavin cofactor (Scheme 4C). Upon light irradiation, the flavin cofactor mediates the first single electron transfer from the enzyme-bound carboxylic acid, initiating the sequence of CO₂ extrusion and back-transfer of the initially abstracted electron to the newly formed C-centered radical [87, 88]. Under light illumination at around 450 nm (blue light), CvFAP efficiently converts long-chain fatty acids into the corresponding alkanes, while achieving high turnover numbers and almost full conversion [89]. An ever growing number of examples highlight the potential of this photoenzyme (and its engineered variants) for various biocatalytic applications, particularly for the synthesis of chiral compounds via kinetic resolution of racemic hydroxyl carboxylic acids [90], or implementation in cascade approaches for the synthesis of secondary alcohols [91] or polymer building blocks [92] from unsaturated fatty acids.

Scheme 4:

Overview on natural photoenzymes. A) Photoreactivation of pyrimidine dimers in UV-damage DNA catalyzed by a photolyase [82]. B) Light-dependent protochlorophyllide reductase (LPOR)-catalyzed reduction of protochlorophyllide to chlorophyllide [84]. C) The fatty acid photodecarboxylase from Chlorella variabilis (CvFAP) catalyzes a decarboxylation reaction upon light illumination, resulting in the formation of hydrocarbons [87]. Adapted with permission from Refs. [12, 16]. Copyright 2022 American Chemical Society.

4.2 Artificial photoenzymes

While the chemistry that can be unlocked from natural photoenzymes is still limited, the combination of biocatalysis and photocatalysis opens up a golden window of opportunities to unlock abiological transformations. Although the concept of using light to harness abiological reaction chemistries from an existing and often engineered enzyme scaffold is a rather new development in photobiocatalysis, a handful of examples already highlight the powerfulness of this approach. The construction of these so-called artificial photoenzymes typically exploits two strategies: (1) the covalent linking of a photocatalysts via an (unnatural) amino acid and using the photocatalyst for direct single-electron transfer; and (2) the conjugation of a photosensitizer nearby an immobilized organometallic complex within a protein scaffold [10, 12, 22]. While artificial enzymes also have been constructed to supply electrons to naturally occurring enzyme cofactors to increase native activity [13, 15, 93], we herein focus on artificial enzymes that have been constructed for non-natural reactivities. Already in 2015, Lewis et al. used a click chemistry approach to conjugate a modified acridinium 9-mesityl-10-methyl (Acr⁺-Mes) cofactor to an unnatural amino acid within a prolyl oligopeptidase to construct an artificial photoenzyme for the sulfoxidation of thioanisoles [94, 95].

Wang et al. constructed an artificial photoenzyme by conjugating a catalytically active organometallic complex in close vicinity to a genetically encoded photoactive chromophore in the superfolder yellow fluorescent protein scaffold [96]. They genetically encoded benzophenone-alanine into the protein scaffold and conjugated a nickel-terpyridine complex for the photocatalytic reduction of CO₂. With the help of protein engineering, the positions of the chromophore and catalyst were tuned, the photosensitizers’ photochemical properties were modulated, and the microchemical environment of the protein was adjusted to enable efficient CO₂ reduction [97]. This photoactive protein scaffold was further modified to enable accomplish cross-couplings of aryl halides via genetically encoded benzophenone chromophore and an adjacent artificial Ni^II(bpy) cofactor [98].

The incorporation of a photosensitizer into a protein scaffold via genetic code expansion for enantioselective [2+2]-cycloadditions via triplet energy transfer has been explored by the groups of Green as well as Chen, Zhong, and Wu [99, 100]. Both groups developed the artificial photoenzyme for [2+2]-cycloadditions based on a different protein scaffold, however, both artificial photobiocatalysts were constructed by genetic code expansion to incorporate a photosensitizer. Upon light irradiation of the photosensitizer, a [2+2]-cycloaddition is promoted. The chosen protein scaffold thereby delivers the needed selectivity, which was further optimized via protein engineering. For instance, the obtained first generation photoenzyme developed by Trimble et al. catalyzed the intramolecular [2+2]-cycloaddition of 4-(but-3-en-1-yloxy)quinolin-2(1H)-one to two regioisomeric products (straight and crossed, Figure 4) [99]. Via directed evolution, an efficient and enantioselective enzyme (up to 99 % ee) promoting selective cycloadditions with >300 turnovers under aerobic conditions was obtained.

Figure 4:

Design of an artificial photenzyme promoting enantioselective [2+2]-cycloadditions via triplet energy transfer (EnT) [99]. The photosensitizer 4-benzoylphenylalanine (BpA) is genetically encoded into the protein scaffold of a computationally designed Diels Alderase, followed by directed evolution, affording an enantioselective photoenzyme for the conversion of 4-(but-3-en-1-yloxy)quinolin-2(1H)-one to the straight or crossed product. Adapted from Ref. [99].

These examples highlight the power of combining photocatalysis with biocatalysis by introducing photocatalysts via bioconjugation with unnatural amino acids or genetically encoding a photosensitizer into a protein scaffold. As such, novel chemical reactivities can be imparted to enzymes, while the design parameters of such biohybrid systems can be modulated to achieve more efficient chemical conversions. However, difficulties arising in the design and construction of artificial photoenzymes impede the chemical diversity that is yet accessible by this approach [12].

4.3 Photoinduced enzyme promiscuity

The protein machineries of living systems are often ‘promiscuous’ – that is, capable of catalyzing reactions other than its biological function. These promiscuous functions can be used to generate catalytic novelty, and promiscuous activities can even be induced or significantly improved by protein engineering [101], [102], [103], [104], [105], [106], [107]. Besides tailoring the substrate specificity or enzyme stability, changing the reaction mechanism of an enzyme is an integral approach that provides access to biocatalysts with a non-biological reaction scope [108, 109].

Recent studies have reported that light can induce catalytic promiscuity in oxidoreductases, which is driven by the excitation of flavin or nicotinamide cofactors as well as organic photosensitizers, leading to the generation of radical intermediates within the enzyme active sites, resulting in various biological transformations. This approach holds great promise for extending the synthetic capabilities of biocatalysts and, when combined with protein engineering, for addressing long-standing selectivity and reactivity challenges in chemical synthesis. This is particularly important as the stereochemical outcome of reactions involving radical intermediates is often difficult to control using existing small molecule catalysts [110].

Recently, Emmanuel et al. have shown that a ketoreductase (KRED) can catalyze the enantioselective dehalogenation of halolactones, a very different reaction from that they were evolved for (Scheme 5A) [111]. In this example, the authors use light to excite the cofactor NADPH bound to the active site of the KRED. Upon light irradiation, the photoexcited cofactor and halolactone substrate generates a charge-transfer complex, leading to the formation of the corresponding cofactor and substrate radicals. The KRED-catalyzed removal of a halogen atom from the halolactone results then in the formation of the dehalogenated chiral lactone product [111].

Scheme 5:

Overview of different light-driven biocatalytic approaches that use light to induce promiscuity in NAD(P)H-dependent enzymes and flavoenzymes. A) Upon light irradiation, a ketoreductase (KRED) catalyzes an enantioselective dehalogenation resulting in the formation of chiral lactones [111]. B) A photoexcited ene-reductase (ERED) catalyzes the radical cyclization of α-chloroamides into γ-lactams [112]. C) Double-bond reductases catalyzing the deacetoxylation of α-acetoxyketones using photoexcited Rose Bengal (RB) [113]. Adapted with permission from Ref. [16].

In follow-up studies, this concept was transferred to flavin-dependent ene-reductases (Scheme 5B) [112]. While intrinsically, these NAD(P)H-dependent enzymes catalyze the two-electron reduction of activated alkenes, the light-induced radical species formation enabled a hydroalkylation used as driving force for non-natural intramolecular cyclization [112], or an intermolecular radical hydroalkylation of terminal alkenes [114]. Unlike to the examples with KREDs, in which NAD(P)H served as the photoactive component, the light-induced single-electron transfer responsible for the promiscuity of ene-reductases is presumably driven by the photoexcitation of the electron-donor complex composed of reduced FMN- and substrate.

These examples rely on the photoinduced electron transfer from NADPH or ground state electron transfer from a flavin hydroquinone. Another approach is to use an exogenous reductant to facilitate electron transfer, while a protein scaffold serves as the chiral catalyst (Scheme 5C) [110, 113]. However, this strategy requires the development of gating strategies to ensure that the formation of the radical species occurs exclusively within the active site of the enzyme. This hypothesis was investigated on a reductive radical deacetoxylation reaction with a NADPH-dependent double bond reductase (DBR) as an enzymatic scaffold. Indeed, the deacetoxylation was observed with good levels of enantioselectivity upon addition of commonly used transition-metal and organic photocatalysts. Moreover, it was proposed that green-light irradiation of RB in the presence of NAD(P)H forms radical RB^●−, an intermediate capable of reducing the enzyme-bond substrate yielding in the formation of a deacetoxylated α-acyl radical. The hydrogen transfer from reduced NADPH yields then the respective product in the final step [113].

5.1 Simultaneous and sequential photochemoenzymatic cascades

The realization of PCE cascades in simultaneous mode is often difficult, mainly because of the often opposite reaction conditions required by the photocatalyst and the enzyme. In addition, stability problems can arise due to the presence of the enzyme, substrate and product in the reaction mixture. Hartwig and colleagues reported a concurrent PCE cascade in which a photocatalyst is used for the isomerization of alkenes, which are further converted by an ERED to generate valuable enantio-enriched products (Scheme 6) [116]. Initially, a number of organometallic and organic photocatalysts were screened for their ability to isomerize the (Z)-configured substrates, and different photocatalysts such as FMN and Ir(III) were identified as most suitable. In the following, the enzymatic reduction of the (Z)-configured substrates was investigated. The cooperative reduction catalyzed by different EREDs and by using different photocatalysts such as FMN and Ir(III) under blue light illumination resulted in the formation of several products with high yields (up to 87 %) and ee’s of >99 %. In this example, two features of the photocatalyst were crucial for performing the cascade in a concurrent reaction mode: (1) the photocatalytic reaction takes place at room temperature to match the enzyme requirements, and (2) the mechanism underlying the photocatalytic reaction involves intermediates that are stable towards water and the functional group in proteins.

Scheme 6:

Concurrent photochemoenzymatic cascade comprising a photocatalytic isomerization with the enzymatic reduction of alkenes to the corresponding enantio-enriched products under blue light illumination [116]. Adapted with permission from Ref. [16].

A second example emphasized the power of combining photocatalysis with biocatalysis for the selective functionalization of (non)-activated C–H bonds, which is an ongoing challenge in classical organic synthesis (Scheme 7) [117]. The applied photocatalyst, sodium anthraquinone sulfonate (SAS), was used for the oxidation of alkanes to the corresponding aldehydes or ketones, followed by the enzymatic transformation of the intermediary aldehyde/ketone to various high-value-added (chiral) products, such as formic esters, lactones, chiral cyanohydrins, chiral acyloins, carboxylic acids, chiral cyclohexanones and amines [117]. With this system, a range of chiral products featuring different functional groups were obtained. Additionally, (R)-benzoin and (R)-mandelonitrile were synthesized at gram-scale with high isolated yields and excellent ee (>99 %), highlighting the applicability of the developed PCE cascade. While the photochemoenzymatic synthesis of these two products was successfully demonstrated at gram-scale, the remaining cascades require further optimization in order to overcome limitations related to catalyst inhibition/deactivation due to the formation of reactive oxygen or radical species and cross-reactivity in the presence of light.

Scheme 7:

Concurrent and step-wise photochemoenzymatic cascades comprising a photocatalytic oxyfunctionalization catalyzed by sodium anthraquinone sulfonate (SAS), followed by further functionalization of the aldehyde or ketone intermediate by different enzymes, such as amine transaminase (ATA), keto reductase (KRED), ene-reductase (ERED), hydroxynitrile lyase (HNL) and benzaldehyde lyase (BAL) [117]. Adapted with permisison from Ref. [16].

In addition to the highly desired functionalization of molecule scaffolds, building more complex molecules from cheap and readily available starting materials is a desirable feature in every synthetic endeavor [118]. The combination of enzymes with photocatalysts thereby offers an attractive approach for implementing multi-step synthetic processes to build molecular complexity. Ravelli and Schmidt recently showed that the merging of photocatalytic C–C bond formation with enzymatic asymmetric reduction enables the direct conversion of simple aldehydes and acrylates or unsaturated carboxylic acids into chiral γ-lactones (Scheme 8) [119]. In this example, the photocatalyst tetrabutylammonium decatungstate (TBADT) is used to catalyze the hydroacylation of the starting olefins, yielding the corresponding keto esters/acids. In the subsequent step, an alcohol dehydrogenase (ADH) converts the keto ester/acid to the corresponding chiral alcohol, which undergoes lactonization to the desired γ-lactone. It was demonstrated that the synthesis of several aliphatic and aromatic γ-lactones was thereby achieved with up to >99 % ee and >99 % yield. This example shows that building molecular complexity from simple, cheap and largely available starting materials is possible by merging photocatalysis with biocatalysis to access high-value added chiral compounds.

Scheme 8:

Step-wise photochemoenzymatic synthesis of chiral γ-lactones comprising decatungstate photocatalyzed hydroacylation of simple olefins with asymmetric reduction of the intermediate 1,4-keto esters/acids catalyzed by ADH. Adapted with permisison from Ref. [119].

Intriguingly, the combination of a photocatalyst with an enzyme can also be used to control the stereochemical outcome of a reaction. For instance, Schmermund et al. showed that the redox potential of a carbon nitride photocatalyst (CN-OA-m) can be tuned by changing the irradiation wavelength to generate electron holes with different oxidation potentials [120]. In combination with an unspecific peroxygenase from A. aegerita, the illumination with of CN-OA-m with green light led to the enantioselective hydroxylation of ethylbenzene to (R)-phenylethanol, whereas blue light irradiation led to the formation of acetophenone, which was further converted with an ADH to the desired (S)-phenylethanol.

5.2 Photoenzymes in biocatalytic cascades

While the majority of PCE cascades rely on the activation of a photocatalytic reaction step by light, also natural and artificial photoenzymes that only perform catalysis upon light irradiation have been implemented in multi-step processes. For instance, a hybrid P450 BM3 biocatalyst containing a covalently attached Ru(II)-diimine photosensitizer has been used in a PCE cascade reaction for the synthesis of trifluoro methylated/hydroxylated substituted arenes [121]. The photochemical properties of the Ru(II) imine photosensitizer allow for the initiation of single electron transfer events, which facilitate the addition of a CF₃ radical to arenes. In the hybrid enzyme, the Ru(II)-diimine photosensitizer provides the necessary electrons to carry out hydroxylation reactions on trifluoromethylated substrates upon visible light activation. While a range of different substrates could be converted, lower yields were obtained in most cases. Beneficially, the regio- and stereoselectivity of the hybrid P450 catalyst enables the differentiation between the trifluoromethylated isomers.

Also ‘true’ photoenzymes have been implemented in cascade approaches (Scheme 9) [89, 91, 122]. For example, Huijbers et al. demonstrated the use of CvFAP in a cascade reaction for the conversion of saturated and unsaturated fatty acids to the corresponding long-chain alkanes or alkenes [89]. This approach for synthesizing long-chain alkenes directly from triglycerides via a two-step cascade reaction represents an interesting alternative strategy to the typical transesterification used in biofuel production. The two-step cascade thereby combines enzymatic hydrolysis of triglycerides with subsequent light-driven decarboxylation catalyzed by CvFAP. In addition to the production of alkanes from triglycerides, another bienzymatic cascade was developed to convert castor oil to (R,Z)-octadec-9-en-7-ol [123]. Therein, non-edible castor oil is hydrolyzed with the help of a lipase, yielding of free ricinoleic acid. Subsequently, CvFAP is catalyzing the decarboxylation under blue light illumination, yielding product concentrations of up to 60 mM [123]. The authors suggested that further optimization of the cascade is needed to compensate for the drop in pH caused by the accumulation of fatty acids in the reaction, and thus achieve even higher product yields [89, 123].

Scheme 9:

Different enzymatic cascades comprising the photodecarboxylase CvFAP [89, 91, 122]. Adapted with permisison from Ref. [16].