The Oxygen Dilemma: A Severe Challenge for the Application of Monooxygenases?

Abstract Monooxygenases are promising catalysts because they in principle enable the organic chemist to perform highly selective oxyfunctionalisation reactions that are otherwise difficult to achieve. For this, monooxygenases require reducing equivalents, to allow reductive activation of molecular oxygen at the enzymes' active sites. However, these reducing equivalents are often delivered to O2 either directly or via a reduced intermediate (uncoupling), yielding hazardous reactive oxygen species and wasting valuable reducing equivalents. The oxygen dilemma arises from monooxygenases' dependency on O2 and the undesired uncoupling reaction. With this contribution we hope to generate a general awareness of the oxygen dilemma and to discuss its nature and some promising solutions.


The Promise of Biocatalytic Oxyfunctionalisation for Organic Synthesis
Oxidoreductases appear set to become practical catalysts for organic synthesis. [1] Following in the path of the well-established hydrolases, [2] the number of industrial [1a, 3] and pre-industrial examples [4] of biocatalytic redox reactions is increasing rapidly. [1c, 5] From as ynthetic point of view monooxygenases are of particular interestf or the organicc hemist because they succeed in balancing high reactivity( needed to activate inert CÀHb onds) and selectivity (by confining the reactive oxygenating speciesi nawell-definedp rotein scaffold). Hence, monooxygenases in principle give access to selective hydroxylations, epoxidations, Baeyer-Villiger oxidations and even halogenations (Scheme1). In comparison with their chemical counterparts, monooxygenases often excel in terms of selectivity and catalystp erformance [turnover numbers (TNs) and turnover frequencies (TOFs)]. [6] Monooxygenases might develop to become practical catalysts, enablingt he organic chemist to perform highly selective oxyfunctionalisation reactions that might otherwise be difficult or requiree xtensive protection group chemistry or longer synthesis routes. There is, however,s tillar ange of issues to be solved to make monooxygenasest ruly practical catalysts. Amongst these there is the oxygen dilemma, whichw eb riefly outline in this contribution.

The Oxygen Dilemma
Am onooxygenase catalyses the selective introductiono fa n activated, electrophilico xygen species (most frequently ah ydroperoxos pecies or ah ighly oxidised transition-metal·oxo complex) into its substrate(s). The activated species is obtained by reductivea ctivation of molecular oxygen att he enzyme's active site.
Monooxygenases are promising catalysts because they in principle enable the organic chemist to perform highlys elective oxyfunctionalisation reactions that are otherwise difficult to achieve. For this, monooxygenases requirer educing equivalents, to allow reductive activation of molecular oxygen at the enzymes' active sites. However,t heser educinge quivalents are often delivered to O 2 either directly or via areduced intermedi-ate (uncoupling), yielding hazardous reactive oxygen species and wasting valuabler educing equivalents. The oxygen dilemma arises from monooxygenases' dependency on O 2 and the undesired uncoupling reaction. With this contribution we hope to generate ag eneral awareness of the oxygen dilemma and to discuss itsnature and somepromising solutions.
One consequence of this catalytic strategy is that monooxygenases need to be supplied with stoichiometrica mountso f reducing equivalents for catalysis. Generally,t hese reducing equivalents are derived from nicotinamide cofactors. These reducing equivalents can "get lost" in different side reactions. In other words, the reducing poweri sd iverted from the target reactiontou nproductive (futile) reduction reactions.I nparticular,O 2 itself is the predominant sink for reducing equivalents and the most important reason for futile side reactions. This phenomenon is well knowninthe scientific literatureasuncoupling (vide infra). Traditionally,t he major issue of uncoupling is believed to be the formation of reactive oxygen species (ROSs),i mpairing enzymes tability and cell viability. Therefore, the majority of in vitro reaction schemes also employ (enzymatic) ROS-scavenging systems such as superoxide dismutases and catalases,w hich remove the different ROSs. This, however, alleviates only one aspectofu ncoupling.
Another aspect of uncoupling,b roached far less frequently, is the futile consumptiono fthe co-substrate (stoichiometric reductant), demanding unnecessary molars urpluses. Once mono oxygenase-catalysed reactions reachp reparative-scale applications, this need for additional enzymes (to destroy the ROSs), as well as for surplus reductants, can significantly impair the attractiveness of these reactions. The dilemma arises from the fact that, although it efficiently fosters the undesired uncoupling, O 2 also cannotbeo mitted from the reaction schemes.

Mechanisms of Uncoupling
Uncoupling occurs mainly 1) duringo xygenation at monooxygenases' active sites, and/or 2) in the delivery of reducing equivalents to thesea ctive sites through electron-transport chains. The main mechanisms are discussed briefly below.B ecause of their predominance as practical catalysts, the discussion is limited to flavin-and haem-dependentm onooxygenases.

Uncoupling within enzymes' active sites
Uncouplingi nt he active sites of flavin-dependent monooxygenases:F lavin monooxygenases make use of a4 a-(hydro)peroxo flavin (obtained after reduction of the enzyme-boundf lavin and subsequentr eaction with molecular oxygen) to oxygenate their substrates. This mostlyr esultsi na4 a-hydroxyflavin, which, after water elimination, enters into an ew catalytic cycle (Scheme 2). The intermediate peroxyflavin species is believed to be stabilised by various interactions, such as by hydrogen bondingt ot he enzyme-bound oxidised nicotinamide cofactor. [7] In some cases, however,t he intermediate 4a-hydroperoxyflavin can also eliminate H 2 O 2 directly,r eturning to the resting state without substrate turnover( step einS cheme 2). [8] The extent of uncouplingv aries both with the typeo ff lavomonooxygenases and with the reactions catalysed by them, as well as with the reaction conditions. [9] Flavoprotein hydroxylases, for example, suffer significantly from uncoupling due to poor stabilityo ft he hydroperoxyflavin.B aeyer-Villiger mono-oxygenases,o nt he other hand, suffer much less from uncoupling (stable peroxyflavin). [10] Underi nitial-rate conditions (low level of conversion of the startingm aterial), "uncoupling generally contributes 5-10 %o f the overall NAD(P)H oxidation rate. Often the product functions as af acilitator for the uncoupling by reversibly binding to the active site and preventing productives ubstrate oxygenation (step ci nS cheme 2). Hence, the uncouplingc ontinuously increases with the reactionprogress. [11] Reversal of the uncoupling reaction (step ei nS cheme 2) as observed with P450 monooxygenases (vide infra) has not yet been observed with flavo-monooxygenases containing the natural cofactors. This is due to the poor electrophilicity of the oxidisedf lavins,r esulting in low rates and unfavourable equilibria for reactions between oxidised flavins and H 2 O 2 .H owever, N-alkylated flavins exhibitingh igherr eactivity have been successfully been integrated into flavin-binding proteins, thereby representing af irst step towards H 2 O 2 -drivenf lavo-monooxygenases. [12] Uncoupling in the active sites of haem-dependent monooxygenases:H aem-dependentm onooxygenases such as the P450 monooxygenases (and non-haem iron monooxygenases) follow as omewhat more complex mechanism than flavo-monooxygenases,(Scheme 3). [13] In most P450-monooxgenase-catalysedr eactions "compound I" (step fi nS cheme 3, formally an Fe V oxo species but more likelya nF e IV ·oxo complex with ad elocalised radical cation within the coordinating porphyrin ligand) represents the oxyfunctionalising species. It is formed through dehydration of an Fe·peroxo complex (step fi nS cheme 3), which itself has been generated by as equence of single-electron transfer steps and oxygen binding( steps a-e in Scheme 3). Analogous-ly to the previouslym entioned flavin-dependent monooxygenases, the Fe·hydroperoxo complex is also known to return to the resting state through H 2 O 2 elimination (uncoupling, step h in Scheme 3). This step is to some extentr eversible and can be exploitedp roductively through the so-called peroxide shunt pathway. [14] The catalytically active compound Ic an also be generated directly from the resting state of aP 450 monooxygenasew itha no rganic peroxide or H 2 O 2 (reversal of step hi n Scheme 3). In principle, this represents am uch simplerr egeneration of P450 monooxygenases (vide infra) but it is very limited by the poor stability of the haem moiety in the presence of even low concentrationsofH 2 O 2 . [15]

Uncoupling in the electron-transport chain
In the previouss ection, uncoupling within monooxygenases' active sites has been discussed. Indeed, with many monooxygenases this also represents the major pathwayo fu ncoupling. This is particularly true for those monooxygenases that utilise NAD(P)H directly as reductant for the prosthetic group. There are, however,a lso al arge number of monooxygenases that obtain their reducing equivalents indirectly:t hat is, through a more or less complicated electron-transport chain (Scheme 4). This sectiond iscusses uncouplingo ccurring within these electron-transport chains.
Why are some mediators inert against O 2 whereas others are not?T he nature of the oxygen dilemma:A si ndicated in Scheme 4, the mediators used to shuttle reducinge quivalents from NAD(P)H to monooxygenases' active sites are O 2 -sensitive whereas NAD(P)H itself is relatively stable towards O 2 .O bviously,t his raises the question of the molecular reason for this discrepancy.
Ap ossible explanation may be found in the spin conservation rule (Wigner's rule). [16] According to this rule, reactions during the course of which the sum of the electron spins changes are "spin-forbidden"a nd therefore slow.S pin-neutral reactions are "spin-allowed" and fast. The electronic ground state of atmospheric oxygen is the so-calledt riplet state ( 3 O 2 ), involving two unpairede lectrons and at otal spin of one (2 1 = 2 ). Hence, electron transfer reactions between 3 O 2 ands ingleelectron mediators are spin-allowed because the sum of spins before and after the reaction does not change (Scheme 5, top). Reactions between 3 O 2 and hydride donors, on the other hand, are spin-forbidden because the sum of spins before and after the reactionc hanges (Scheme 5, bottom). [17] This behaviour can nicely be exemplified with the organometallica rtificial mediator [Cp*Rh(bpy)(H 2 O)] 2 + (and itsr educed form [Cp*Rh(bpy)H] + ). Depending on the reaction conditions, this acts either as as ingle-electron mediatoro ra sa hydride-transfer mediator ( Figure 1). [18] Whereast he reduction of NAD + (hydride transfer reaction) was not influencedb yt he presence or absence of O 2 the reduction of cytochrome C (single-electron transfer)p roceeded at significant rates only after the dissolved molecular oxygen had been depleted.
Hence, NAD(P)H (acting as ah ydride-transfer mediator) is metastable in the presence of atmospherico xygen whereas reduced ferredoxins (acting as SET mediators) and reduced flavins react readily.This constitutes the oxygen dilemma.
Scheme3.Simplified mechanism of haem-dependentm onooxygenases, consisting of two single-electron transfer steps to haem iron (steps ba nd d) and formation of "compound I" (step f) to perform the oxygenation reaction (step g). The elimination of H 2 O 2 from the intermediate Fe III peroxo species is showni nstep h.
Scheme5.Molecular oxygen (two spins)can react quickly with single-electron mediatorsbecause the sum of spins does not change during the reaction (spin-allowed reaction). Hydride donors (or two-electronm ediators in general) react with molecular oxygen far more slowly becauset he sum of spins changes during the reaction (spin-forbidden reaction).
Scheme4.Simplified molecular architecture of multicomponent monooxygenases that are not directly dependento nN AD(P)H. Blue:path of reducing equivalents. NAD(P)H serves as ag eneral reductant, transferring its reducing equivalents to amediator molecule (either aflavinora ni ron-sulfurc luster protein) with catalysis by ar eductase. The usually protein-based mediator delivers the reducinge quivalents to the monooxygenase subunit for productiveoxygen activation. However, direct reaction of the reducedm ediator with O 2 leads to futiler eoxidation and (eventually) H 2 O 2 formation. Here it is worth briefly discussing the mechanism of flavin oxidation. Reduced flavins (FADH 2 ,F MNH 2 ,R fH 2 )a re generally known to be very reactive with molecularo xygen. Hence, significant uncoupling can be expectedi nc ases of those monooxygenases that dependo nf reely diffusing reduced flavins as mediators.
Frequently,areoxidation mechanism in the style of the enzymaticu ncoupling mechanism (Scheme 2) is proposed. However,a se arly as the 1970s, Massey and co-workers intensively investigated the redox chemistry of flavins and demonstrated that, in solution,aradical-based reoxidation mechanism most probably prevails [Eqs.
k 2~2 60 s À k 4~~1 10 4 -1 10 6 m À1 s À (depending on pH) k 5~1 10 6 m À1 s À k 6~8 10 7 m À1 s À Flavin-dependent monooxygenases can be classified according to the electrond onor providing the reducing equivalents neededf or the reductivea ctivation of molecular oxygen (Table 1). [8b, 9a, c] Groups Aa nd Bc onsist of enzymest hat rely on NAD(P)H as external electron donor.G roupsC -F are two-protein systems, composed in each case of am onooxygenase and af lavin reductase.G roups Ga nd Hc ontain internal monooxygenases that reducet he flavin cofactor through substrate oxidation. With respect to the oxygen dilemma, groups A, B, Ga nd H, on the one hand, and groups C-F,o nt he other,s hould be distinguished. The first ones (A, B, G, H) rely on O 2 -stable reductants whereas the reductantso ft he latter series (C-F) are highlyr eactive with O 2 .H ence, especially for the latter set of enzymes, uncoupling not only occurs within the active sites (as discussed above) but also in the electron-transport chain.
The molecular reason for the (seemingly unnecessarily) more complicated electron-transport chains in the cases of some flavin-dependent monooxygenases [21] remains mysterious, especially when the additional loss of valuable reducing equivalents is considered.
P450 monooxygenases:I nt he case of haem-dependent monooxygenases the complicated electron-transport chain is amechanistic requirement:N AD(P)H serves exclusively as hydride donorw hereas the mechanismo fh aem monooxygenases involvest wo individual SET steps. Therefore, arelay system trans- forming ah ydride-transfer step into two sequential single-electron-transfer steps is necessary to link P450 monooxygenases to the microbial[NAD(P)H] energy metabolism.P450 monooxygenases can be classified according to the molecular architecture of the electron-transport chain into one-, two-and threecomponent P450 monooxygenases( Scheme 6). [22] They all have in commont hat af lavin-dependent (FAD or FMN) reductase catalysest he initial oxidation of NAD(P)H and enables two successive SET steps. In the cases of the one-and two-component P450 monooxygenases these SETso ccur directly to the monooxygenase subunit, whereas in that of the three-component systems the reductase reduces af reely diffusing ferredoxin. The reduced ferredoxin then delivers two electrons in two SETst ot he monooxygenase subunit. Ap rototypeo fb acterial P450 systems is the P450cam system from Pseudomonas putida,inwhich the cytochrome catalyses the hydroxylation reaction. Furthermore, the system incorporates aF AD-containing reductase as well as an iron-sulfur protein (putidaredoxin). This [2 Fe-2 S] ferredoxin plays the role of an electron shuttle, transferring the two electrons one at at ime from putidaredoxin reductaset oP 450cam. Therefore, the putidaredoxin can be regardeda sanaturalm ediatori nt his three-component system. Therefore, it can be expected that the two-and three-component P450 monooxygenases (in which electron supply relies on diffusionale lectron mediators) in particular should be especially prone to the oxygen dilemma. The resulting uncoupling is well known in the literature and has been discussed in several review articles. [23] Fasan recently pointed out that for many P450 monooxygenases the coupling efficiency with the natural substrate(s) can be as high as 90-98 % [23b] but can decrease significantly to almost any value ranging from < 1% to 30-40 % in the presence of unnatural substrates. It mightb ep ossible to overcome this issue by improving the interaction between the haem moiety and the mediator. The electron transfer between the electron-transport proteins and the monooxygenase subunit is essential and has to be fine-tuned to allow for efficient (that is, productive) oxygen activation and product formation. [24] The P450 monooxygenase from Bacillus megaterium (P450 BM3) constituted the first example of as elf-sufficient P450 monooxygenase, gathering all electron-transport components in one single polypeptide. [25] The sometimes very high coupling efficiencies observed in P450 BM3 have inspired researcherstoconstruct artificial fusion proteins. Some successful examples of this strategyh ave been summarised by Hlavica. [26] However, it should also be mentioned heret hat this approach does not always lead to improved self-sufficient enzymes. Even in P450 BM3 the uncoupling can be as high as 60 %o re ven more. [27] Overall, it can be concluded that uncoupling in monooxygenases occurs naturally. In the case of haem-dependentm onooxygenases in which single-electron transfers are am echanistic necessity,u ncoupling is even inevitable.T his raises the question of why natural oxygenation schemes would rely on such wasteful and dangerousp rocesses.O ne possible explanation might be that the uncoupling generated by monooxygenases (and their electron-transport chains) carries no weighti nc omparisonw ith mitochondrial uncoupling [28] or the uncoupling occurring in chloroplasts. [29] Hence, the oxygen dilemma might be just the price to pay for an aerobic life.

Ways to Deal with the Oxygen Dilemma
Essentially,t he oxygen dilemmai sakinetic phenomenon. Reducinge quivalents can either be deliveredt ot he monooxygenases productively or they can be diverted to free molecular oxygen, therebyw asting valuable reducing equivalents and yielding ROSs. Approaches to alleviation of the oxygen dilemma should therefore focus on increasing the rate of the first (desired) pathway while slowingd own the latter (undesired) reaction.
To assuage uncoupling within the monooxygenases' active sites, protein engineering appearst ob et he methodo fc hoice. Especially for the flavo-monooxygenases, structural information about the stabilisation of the reactive 4a-(hydro)peroxyflavin is available [7a, b, 30] and can serve as starting point to improve the efficiency of other flavo-monooxygenases. The negative effect of accumulating product on the uncoupling can best be addressed by in situ product removal strategies, maintaining the product concentration in the aqueous reaction mixturel ow and facilitating downstream processing. [11,31] Improving the coupling efficiency of the electron-transport chainsi nm ultiple-component monooxygenases can be achieved by improving the interaction betweent he reduced natural mediators and the monooxygenase subunit. In the case of P450 monooxygenases the concept of molecular building blocks-that is, combining monooxygenases with the most suitable mediators and reductasesf or efficient interaction and www.chembiochem.org electron transfer-might represent af uture approach. [24] Similar trials with two-component flavo-monooxygenase have been reported. [32] Another line of research involves simplified electron-transport chains. Ideally,o ne single catalystf unctions as relay system between as acrificial electron donor and am onooxygenase (Scheme 7), therebys ubstituting the nicotinamide cofactor together with the corresponding reductasesa nd mediators. The promise of this approach lies in its highd egree of simplification and, hopefully,g reater robustness of this reaction scheme.
The most common regenerationc atalysts are transitionmetal complexes such as cobalts epulchrate, [33] cobaltocene, [34] or ruthenium [35] or rhodiumcomplexes. [21d, 36] Organicmediators such as flavins [37] or redox-active dyes [38] have also been reported. Unfortunately,l ittle information about the coupling efficiency can be found in most of these contributions. However, from the few cases in whicht he uncoupling is quantified it becomes clear that classical one-electronm ediators are rather poor in their coupling efficiency whereas hydride donors are usually significantly better. An additional complication occurs in the case of electrochemicalr eaction systems. [36b, 37e, 39] Generally,t he cathode potentials required for efficient reduction of the mediator (natural or artificial) used are more negative than the O 2 reduction potential. Hence, direct cathodic O 2 reduction occurs during the electrolyses, reducing the current yield (significantly contributing to uncoupling) and generating ROSs.
Severala pproaches to dealing with the oxygen dilemma in artificial regenerations ystemsh ave been proposed. Of these, af ew (promising) strategies are discussed below.
Cheruzel and co-workers have reported covalentl inkages between P450 monooxygenases and (mostly Ru-based) photosensitisers/mediators with great improvementi nt he reaction rates. Possibly,t his can also be attributed-at least to some extent-to decreased futile reoxidation of the reduced mediators due to elimination of diffusion limitations (Scheme 8). [35a, c, 40] The previously mentioned direct cathodic reduction of O 2 can in principle be alleviated by choice of as uitable reactor design minimising the O 2 concentration in the cathode compartment. [37d, 39a, b, 41] Similarly,e ngineering the monooxygenase mediatori nteraction on the basis of rational design might improve the efficiency of electron transfer. [33e, 42] Another possible handle could be the choice of electrode material. Ar ecent study,f or example, demonstrated that glassy carbon exhibited significantly higher reduction rates with Co sepulchrate than Pt electrodes, together with reduced O 2 reduction rates. [43] Employing O 2 -stable artificial mediators could be aviable approach.F or example, deazaflavins, whichh ave been known for decades as O 2 -resistant analogues of the common flavin mediators, [19,44] have been used as photocatalysts/mediators to regenerate P450 BM3a nd Old Yellow Enzymes. In both cases O 2 efficiently interfered in the reduction of the enzyme prosthetic group (haem or FMN). Changing the mediator from an ormal flavin to ad eazaflavin analoguei ncreasedt he efficiency of the reaction schemes significantly (Scheme 9). [37b, 45] Finally,i ti sa lso worth mentioning that "exploiting" the oxygen dilemma by using H 2 O 2 -dependent enzymes is av iable and very promising solution (Scheme 10).
Peroxygenases utilise H 2 O 2 to generate the catalytically active compound I, known from P450 chemistry (reversal of step hi nS cheme 3), directly. [46] Fortunately,t he number of syntheticallyi nteresting peroxygenases is increasingr apidly, [47] and new protein engineering tools to adjust the substrate scope and selectivity of these enzymes are available. [48] The issue of poor robustness of these haem-dependent enzymes against H 2 O 2 can be addressedb yv ariousc atalytic in situ H 2 O 2 generation systems. [49] Therefore, exciting new biocatalytic oxyfunctionalisationr eactions can be expected in the near future.
Scheme7.Simplified regeneration of (mono)oxygenases by directr eductive regeneration of the enzymes' active sites.

Conclusions
The oxygen dilemma is ar eality that has to be faced in biocatalytic oxyfunctionalisation chemistry.T he high reactivity of molecular (triplet)o xygen with the most common natural and man-made redox mediators interferes in most electron-transport chains delivering electrons to monooxygenases. To day, the major challenge of this well-known uncoupling is seen in the formation of reactiveo xygen species (ROSs) andt heir negative influence on the robustnesso ft he (bio)catalysts used. This issue is relativelye asily alleviated by using nature's arsenal of ROS-scavenging enzymes (such as superoxide dismutases and catalases).
In view of the fact that monooxygenases today are mostly used in the fine chemicals and pharmaceutical sectors, the wasteful nature of uncoupling is of lesser importance. [50] If large-scale applications of monooxygenases are envisioned, however,t he waste of valuabler eduction equivalents comes to the fore. From the economic point of view,b ulk products with lower margins do not allow for wastefulp rocesses. Also from an environmental point of view,t hough, futile waste of resources has to be avoided. [51] Af irst step towards more redox-efficient processes should be aq uantitative understanding of the uncoupling. More experimental data quantifying the coupling efficiency of monooxygenase reactions( not only from initial rate experimentsb ut also throughout the processes)s hould be helpful in creating ag enerala wareness of the oxygen dilemma.I nt his contribution af ew promising approaches have been mentioned. We hope that this minireview might encourage other researchers to focus on solving the oxygen dilemma to makeb iocatalytic oxyfunctionalisation truly efficient and practical.