A surprising observation that oxygen can affect the product enantiopurity of an enzyme-catalysed reaction



N. S. Scrutton, Manchester Interdisciplinary Biocentre and Faculty of Life Sciences, University of Manchester, Manchester M1 7DN, UK

Fax: +44 161 3068918

Tel: +44 161 3065152

E-mail: nigel.scrutton@manchester.ac.uk


Enzymes are natural catalysts, controlling reactions with typically high stereospecificity and enantiospecificity in substrate selection and/or product formation. This makes them useful in the synthesis of industrially relevant compounds, particularly where highly enantiopure products are required. The flavoprotein pentaerythritol tetranitrate (PETN) reductase is a member of the Old Yellow Enzyme family, and catalyses the asymmetric reduction of β-alkyl-β-arylnitroalkenes. Under aerobic conditions, it additionally undergoes futile cycles of NAD(P)H reduction of flavin, followed by reoxidation by oxygen, which generates the reactive oxygen species (ROS) hydrogen peroxide and superoxide. Prior studies have shown that not all reactions catalysed by PETN reductase yield enantiopure products, such as the reduction of (E)-2-phenyl-1-nitroprop-1-ene (PNE) to produce (S)-2-phenyl-1-nitropropane (PNA) with variable enantiomeric excess (ee). Recent independent studies of (E)-PNE reduction by PETN reductase showed that the major product formed could be switched to (R)-PNA, depending on the reaction conditions. We investigated this phenomenon, and found that the presence of oxygen and ROS influenced the overall product enantiopurity. Anaerobic reactions produced consistently higher nitroalkane (S)-PNA product yields than aerobic reactions (64% versus 28%). The presence of oxygen dramatically increased the preference for (R)-PNA formation (up to 52% ee). Conversely, the presence of the ROS superoxide and hydrogen peroxide switched the preference to (S)-PNA product formation. Given that oxygen has no role in the natural catalytic cycle, these findings demonstrate a remarkable ability to manipulate product enantiopurity of this enzyme-catalysed reaction by simple manipulation of reaction conditions. Potential mechanisms of this unusual behaviour are discussed.




enatiomeric excess


oxidized flavin


reduced flavin


glucose dehydrogenase


horseradish peroxidase


Old Yellow Enzyme


pentaerythritol tetranitrate






reactive oxygen species


superoxide dismutase


xanthine oxidase


The Old Yellow Enzyme (OYE) family of FMN-containing oxidoreductases [1, 2] catalyses the NAD(P)H-dependent asymmetric reduction of the C=C bonds of a variety of industrially useful α,β-unsaturated aldehydes, ketones, and nitroalkenes [3-8]. These reactions proceed in two stages: (a) NAD(P)H oxidation by hydride transfer to the FMN cofactor (reductive half-reaction); and (b) reduction of activated alkenes by hydride transfer from the reduced flavin (Flred), with a net trans-hydrogenation (oxidative half-reaction; Scheme 1) [9, 10]. OYEs also possesses NAD(P)H oxidase activity, whereby oxygen reoxidizes reduced FMN in the absence of activated alkene substrates; however, this oxidase activity is not part of the natural catalytic cycle. Kinetic investigations are usually performed under strictly anaerobic conditions to eliminate this competing activity. One extensively studied OYE family member is pentaerythritol tetranitrate (PETN) reductase (EC from Enterobacter cloacae PB2 [11]. This ‘ene’-reductase was originally isolated because of its ability to degrade a variety of high explosives, such as nitroaromatics and nitrate esters [12, 13]. It was later found to reduce the C=C bonds of a variety of α,β-unsaturated aldehydes, ketones, carboxylic acids, nitroalkenes, and steroids [14-16], which is consistent with the reported activities of other OYEs.

Asymmetric hydrogenation reactions are important to industry, as up to two new stereogenic centres can be generated [2]. Recent reviews have highlighted the biocatalytic potential of the OYE family in performing asymmetric activated alkene reduction [2, 17]. The presence of oxygen is known to reduce the overall steady-state rate of PETN reductase-catalysed alkene reduction, owing to futile cycles of NADPH-dependent reduction of flavin and reoxidation by oxygen; oxygen-mediated oxidation of the flavin competes with alkene substrate reduction by enzyme-bound FMNH2[14]. OYE-catalysed reactions performed aerobically are known to generate reactive oxygen species (ROS). In spite of this, most reported OYE-catalysed biocatalytic investigations have been performed aerobically, perhaps because of perceived difficulties in performing anaerobic bioreductions in large-scale reactions.

Enzymes catalyse reactions with often unmatched high efficiency, chemoselectivity, regioselectivity, and stereoselectivity [18], but not all generate enantiopure products. For example, PETN reductase-catalysed anaerobic reduction of (Z)-2-phenyl-1-nitroprop-1-ene (PNE) generates near enantiopure (S)-2-phenyl-1-nitropropane (PNA), but equivalent reactions with the (E)-isomers of PNE generate (S)-PNA with a lower enantiomeric excess (ee) (~ 66–89%) [16, 19]. Reaction optimization techniques, such as varying reaction time and enzyme concentration, have been shown to increase product enantiopurity in some cases [16, 20]. In contrast to these observations, a recent investigation by Mueller et al. into the reduction of (E)-PNE by PETN reductase under aerobic conditions showed a surprising switch in enantiospecificity by producing predominantly (R)-PNA (57% ee) [21]. Other parameters, such as buffer identity and pH, differed between the two reaction protocols [14, 16, 21]. We have investigated this phenomenon by performing comparative reactions under a variety of reaction conditions to determine the likely cause(s) of this apparent inconsistency in the enantioselectivity of (E)-PNE reduction. Our studies revealed that oxygen and ROS have opposing effects in controlling the enantioselective outcome of (E)-PNE reduction by PETN reductase. This remarkable influence of oxygen and ROS on the outcome of the PETN reductase-catalysed reduction of (E)-PNE highlights the need to consider the influence of oxygen and ROS when using OYEs in aerobic biotransformation reactions. Moreover, our studies highlight how the presence of a surrogate substrate (oxygen), or its reaction products (ROS), can influence the stereoselectivity of (E)-PNE reduction without becoming directly involved in the redox chemistry of (E)-PNE reduction.

Results and Discussion

Comparative reactions

Investigations into the effect of reaction conditions on product enantiopurity cannot be performed with traditional steady-state or fast reaction kinetics, as the individual product enantiomers are indistinguishable by UV–visible and fluorescence spectroscopy. Therefore, asymmetric bioreductions of (E)-PNE were performed under multiple turnover conditions, where the product yield and percentage ee were determined with (a)chiral GC [16]. The reactions consist of buffer (50 mm), alkene (E)-PNE (5 mm), NAD(P)H (6–15 mm), and enzyme (2 μm). Reactions were incubated for 24 h at 30 °C, and this was followed by the determination of product yield (percentage product formed), enantiomer identification [(R) versus (S)] and ee (expressed as percentage excess of one enantiomer over the other).

Comparative analysis of the reaction conditions for the two independent studies [16, 21] of PETN reductase-catalysed reduction of (E)-PNE showed that a number of parameters differed between the protocols. They consisted of: (a) enzyme sequence (native versus C-terminally His8-tagged); (b) buffer type and pH [50 mm Tris (pH 7.5) versus 50 mm KH2PO4/K2HPO4 (pH 7.0)]; (c) coenzyme used (NADH versus NADPH); and (d) oxygen content (aerobic versus anaerobic). We performed comparative biotransformations of PETN reductase-catalysed (E)-PNE reduction, where all four differing parameters were varied (Table 1). These results showed that the product yield and ee were independent of the presence or absence of the His8-tag. This is in agreement with previous steady-state spectroscopic kinetic studies, which showed that the reductive and oxidative half-reactions were essentially unchanged between these two enzyme forms [22].

Table 1. Influence of reaction conditions on the reduction of (E)-PNE by PETN reductase under aerobic and anaerobic conditions. Reactions (1 mL) were performed aerobically in buffer (50 mm) containing alkene (5 mm), PETN reductase (2 μm), and NAD(P)H (6 mm). Reactions were agitated at 30 °C for 45 h at 130 r.p.m. PO4, phosphate buffer; His8-tag, His8-tagged wild-type enzyme. Conversion and yield were determined by GC with a DB-Wax column. ee was determined by GC
EnzymeBuffer (pH)CofactorO2/N2Conversion (%)Yield (%)ee (%)Configuration
Wild typePO4 (7.0)NADPHN2928558(S)
Tris (7.5)NADPH918559(S)
PO4 (7.0)NADPHO2755128(S)
Tris (7.5)NADPH796223(S)
His8-tagPO4 (7.0)NADPHN2868159(S)
Tris (7.5)NADPH908559(S)
PO4 (7.0)NADPHO2635836(S)
Tris (7.5)NADPH815327(S)

There was a moderate decrease in product yield in aerobic reactions, presumably because of competition between (E)-PNE and oxygen for the reducing equivalents in FMNH2[17]. This has been seen previously in comparative aerobic and anaerobic steady-state kinetic studies of PETN reductase-catalysed reduction of 2-cyclohexen-1-one and 2-methyl-pentenal [14]. Additionally, there was a dramatic decrease in (S)-PNA enantiopurity (64% versus 24% ee) under aerobic conditions. Interestingly, only aerobic reactions in the presence of NADH and Tris buffer showed a predominance of (R)-PNA formation, albeit with low enantiopurity (18–21%; Table 1). These latter conditions closely match the experimental protocols of Mueller et al. [21], who demonstrated predominantly (R)-PNA formation. The differences in conversion and product yields were attributable to the production of 5–10% of a byproduct, 2-phenylpropanal oxime, as seen in previous studies, in addition to 2-phenylpropanal and 2-phenylpropan-1-ol [20, 22, 23]. The mechanism of byproduct formation is unclear, and two contrasting mechanisms have been proposed previously [20, 23, 24].

To determine which reaction parameter(s) had a significant effect on the product enantiopreference of PETN reductase-catalysed (E)-PNE reduction, we performed further comparative studies in which one or more of the reaction conditions were varied. The following variables were studied: (a) buffer type; (b) pH; (c) coenzyme type; (d) enzyme concentration; (e) reaction time; and (f) oxygen/ROS content. The experimental data and reaction conditions can be found in Tables S1–S7.

Reduction of (E)-PNE by PETN reductase in the presence of five different buffers (phosphate, Tris, Mops, Hepes, and Pipes) at pH 7.0 showed consistent (S)-PNA product yields and ee (Table S1). These values increased slightly under anaerobic conditions. When the buffer composition was maintained but the pH of the reaction was varied (BisTris/Hepes/Ches; pH 5.0–9.0), the ee for (S)-PNA remained approximately constant, whether reactions were performed aerobically or anaerobically (Table S2). The yields varied with pH (PETN reductase activity is pH-dependent), with the optimum pH being 7.5. When the effects of both pH and coenzyme type (NADH versus NADPH) under aerobic reactions were compared (Table S3), approximately constant product ees and pH-dependent variation in product yields were observed. Combined, these results show that buffer identity and pH do not significantly influence the overall product ee, but that pH and oxygen content do affect product yield. Also, aerobic conditions lower the product ee.

Further comparative reactions were performed in phosphate buffer containing NAD(P)H with variable enzyme concentrations (2 nm to 20 μm) under aerobic and anaerobic conditions (Tables S4 and S5). The yields decreased with decreasing enzyme concentration, and the (S)-PNA ee also decreased to near racemic levels. Similar reactions performed aerobically in Tris buffer showed a switch to predominantly (R)-PNA formation, with low ee values (9–26%). Therefore, the enzyme concentration has some influence on the overall product ee, but only in the presence of oxygen.

There was evidence from nonenzymatic control reactions of significant (E)-PNE and/or product decomposition (Table S6), so shorter reaction times are preferable with this substrate. We monitored the anaerobic/aerobic reduction of (E)-PNE with NAD(P)H over 24 h (Tables S6 and S7) to determine whether the reaction time impacted on product enantiopurity. As expected, anaerobic reactions were consistent in terms of (S)-PNA product yield and ee, and had higher yields than the aerobic reactions (Table S7). Interestingly, when these reactions were performed with the (Z)-isomer of PNE, oxygen had no significant impact on the reaction (Fig. 1).

Figure 1.

Effect of oxygen levels and cofactor type on the product enantiopurity in the reduction of (Z)-PNE and (E)-PNE by PETN reductase. Reactions (1 mL) were performed aerobically or anaerobically in Tris buffer containing alkene (5 mm), PETN reductase (2 μm), and NAD(P)H (10 mm). Reactions were agitated at 30 °C for 0.5–24 h at 130 r.p.m. Open circles: reactions with (Z)-PNE and NADPH. Closed circles: reactions with (E)-PNE and NADPH. Open triangles: reactions with (Z)-PNE and NADH. Closed triangles: reactions with (E)-PNE and NADH. Solid line: anaerobic reactions. Dashed line: aerobic reactions.

Remarkably, time-dependent differences in product enantiopurity were seen between PETN reductase-catalysed reductions of (E)-PNE in the presence of oxygen (Fig. 1; Table S7). There was a switch from the formation of mostly (R)-PNA to the production of increasing levels of (S)-PNA with increasing time during aerobic biotransformations of (E)-PNE. This was independent of coenzyme identity, and the final product ee was lower than in anaerobic reactions (16–23% versus 54–56%). The timing of this switch in enantiopreference from (R)-PNA to (S)-PNA was variable, but longer reaction times produced more (S)-PNA (Fig. 1). These data show clearly that the product enantiomeric identity and purity are strongly influenced by the oxygen content of the reaction, but variations in the reaction time, coenzyme/buffer identity and enzyme concentration have less effect.

Detection of ROS

We have shown that aerobic reaction conditions were necessary to increase the level of (R)-PNA production over (S)-PNA production. Previous studies by others have shown that the reduced flavin (Flred) of several OYE family members reacts with oxygen to generate the ROS hydrogen peroxide and superoxide (Scheme 2) [25, 26]. These ROS form by an initial electron transfer from Flred, forming a caged radical pair of flavin neutral semiquinone and superoxide anion (Scheme 2) [25]. This complex then progresses along one of three reaction pathways leading to the formation of oxidized flavin (Flox). In the first pathway, a C4a peroxyflavin species is formed, which, following protonation, dissociates to form Flox with release of hydrogen peroxide. Alternatively, dissociation of the superoxide anion yields a free flavin semiquinone (second pathway). This is followed by oxygen binding, electron transfer and superoxide anion release to generate Flox. The third pathway involves spontaneous dismutation of the superoxide and flavin radicals to produce hydrogen peroxide and Flox[25].

Aerobic biotransformation reactions catalysed by OYEs can potentially form one or more ROS defined by the pathways of Scheme 2. It is therefore possible that the altered stereochemical outcome observed in aerobic biotransformation reactions with (E)-PNE may be attributable to the presence of (a) oxygen alone, (b) one or both ROS, or (c) a combination of oxygen and ROS. To investigate this, additional aerobic reactions were performed in the presence of the enzymes superoxide dismutase (SOD) (Reaction 1) and/or catalase (Reaction 2) to eliminate superoxide and hydrogen peroxide, respectively [25].

display math( Reaction 1)
display math(Reaction 2)

Initially, we needed to demonstrate whether PETN reductase generates hydrogen peroxide and/or superoxide during aerobic reactions with NAD(P)H in the absence of an additional hydride acceptor (Table 2). In the case of superoxide formation, we continuously monitored the reduction of cytochrome c by superoxide at 550 nm until PETN reductase had consumed all of the NAD(P)H [25]. To determine hydrogen peroxide production, NAD(P)H (20 μm) consumption was monitored (340 nm) until it was depleted. A colorimetric reaction was then performed with a 3,3′-diaminobenzidine (DAB)/gelatin solution containing the enzyme horseradish peroxidase (HRP) [27]. The final colour development was measured at 465 nm.

Table 2. Detection of superoxide and hydrogen peroxide production by PETN reductase-catalysed reduction of 20 μm NAD(P)H under aerobic conditions. Reactions (1.0 mL) were performed in buffer (50 mm phosphate or Tris), NAD(P)H (20 μm) and PETN reductase (0.5 μm) ± cytochrome c (20 μm) in the presence or absence of excess SOD (20 U) and/or catalase (20 U). Further details of the reaction protocols are given in 'Experimental procedures'
BufferROS eliminatorSuperoxide yield (μm)H2O2 yield (μm)
  1. a

    Data for the endpoint assays in the absence of any SOD or catalase have been corrected for background levels. Backgrounds for superoxide and hydrogen peroxide yields were determined by the addition of SOD/catalase and catalase alone, respectively.

Phosphate, pH 7.0Nonea7.1 ± 0.023.9 ± 0.28.8 ± 0.15.2 ± 0.2
 Catalase7.7 ± 0.111.5 ± 0.61.6 ± 0.21.6 ± 0.2
 SOD/catalase0.8 ± 0.16.2 ± 0.3
Tris, pH 7.5Nonea5.5 ± 0.26.0 ± 0.36.4 ± 0.25.3 ± 0.1
 Catalase6.9 ± 0.110.4 ± 0.51.5 ± 0.21.5 ± 0.2
 SOD/catalase1.0 ± 0.13.7 ± 0.2

The yields of superoxide varied from 3.9 to 7.1 micromole per 20 micromole NAD(P)H reduced with the two buffer conditions (Table 2), after correction for background rates. However, owing to the inherent instability of superoxide, these values are likely to be an underestimation. The presence of hydrogen peroxide in the reaction is likely to further reduce the overall superoxide yield detected, owing to its ability to reoxidize reduced cytochrome c[25]. Additionally, superoxide is known to reoxidize Flred at a higher rate than oxygen [26]. The background rates of apparent superoxide production (SOD/catalase reactions) are most likely attributable to a small proportion of cytochrome c reduction, caused by a direct interaction with Flred[25]. In the case of hydrogen peroxide production, the yields varied from 5.2 to 8.8 micromole per 20 micromole NAD(P)H (Table 2). The background levels in this case (catalase reactions) may be attributable to a minor interaction of another ROS with the DAB reagent.

Comparative kinetic studies of the oxidation of NAD(P)H by PETN reductase were performed in the presence or absence of SOD and catalase to determine whether the ROS eliminators affect the overall reaction rate (Table 3). Reactions were performed in the presence or absence of a competing hydride acceptor (100 μm ketoisophorone, an alternative substrate of PETN reductase [14]). There were only minor differences in rate between reactions performed in the presence or absence of SOD/catalase, and this effect was essentially absent in Tris buffer reactions. Reactions tended to be slightly slower in the absence of superoxide and hydrogen peroxide (SOD/catalase reactions), as superoxide is known to rapidly reoxidize Flred.

Table 3. Detection of superoxide and hydrogen peroxide production by PETN reductase-catalysed reduction of 20 μm NAD(P)H under aerobic conditions. Reactions (1.0 mL) were performed in buffer (50 mm phosphate or Tris), NAD(P)H (100 μm) ± ketoisophorone (100 μm) and PETN reductase (15–500 nm) in the presence or absence of excess SOD (20 U) and catalase (20 U). Reactions were monitored at 340 nm for 2 min at 25 °C
BufferROS eliminatorRate with alkene (s−1)Rate (no alkene) (s−1)
Phosphate, pH 7.0None6.20 ± 0.110.08 ± 0.010.22 ± 0.010.10 ± 0.01
SOD/catalase5.95 ± 0.120.08 ± 0.010.13 ± 0.010.07 ± 0.02
Tris, pH 7.5None5.58 ± 0.220.07 ± 0.010.14 ± 0.010.07 ± 0.01
SOD/catalase5.40 ± 0.040.07 ± 0.010.14 ± 0.010.06 ± 0.01

Effect of ROS on biotransformations

The effect of ROS on product enantiopurity was investigated in the reduction of (E)-PNE by PETN reductase under the following conditions: (a) anaerobic (< 5 p.p.m. oxygen species); (b) standard aerobic conditions (contains oxygen, superoxide, and hydrogen peroxide); (c) aerobic + SOD (contains oxygen and hydrogen peroxide); (d) aerobic + catalase (contains oxygen and superoxide); and (e) aerobic + SOD + catalase (contains only oxygen). Control reactions were performed in the absence of PETN reductase, and the presence of significant SOD and catalase activity was detected after 24 h (results not shown; see 'Experimental procedures' for details).

Comparative reactions gave a variable 42–68% yield of the alkane PNA (Table S8), whereas oxime yields were more similar (11–22%). No significant aldehyde byproduct was detected in any reaction [16]. Reactions with oxygen, but with no ROS present (SOD/catalase), generated the highest ee of (R)-PNA overall (43–52%), except in the presence of NADPH and Tris buffer (Fig. 2). The presence of SOD (i.e. only oxygen and hydrogen peroxide present) led to significant (R)-PNA formation, whereas the presence of catalase (i.e. only oxygen and superoxide present) led to near racemic products. However, the effect of hydrogen peroxide on reactions in the presence of SOD is magnified, as hydrogen peroxide is produced by both PETN reductase (Scheme 2) and SOD (Reaction 1). Reactions were also performed in the presence of catalytic concentrations of NADP+, employing a glucose/glucose dehydrogenase (GDH) recycling system to regenerate NADPH, to ensure that the presence of excess NAD(P)H does not impact on the stereochemical outcome of the reactions. Similar trends in enantiomeric preference were obtained under these conditions, suggesting that the NAD(P)H concentration does not significantly affect the product stereochemistry (results not shown).

Figure 2.

Effect of ROS eliminators on the product enantiopurity in the reduction of (E)-PNE by PETN reductase. Reactions (1 mL) were performed aerobically in buffer (phosphate or Tris), (E)-PNE (5 mm), PETN reductase (2 μm) and NADPH (15 mm) in the presence or absence of SOD (20 U) and/or catalase (20 U). Reactions were agitated at 30 °C for 24 h at 130 r.p.m. Downward slanting lines: anaerobic reactions. Upward slanting lines: contains oxygen (SOD/catalase). Crosses: contains oxygen and hydrogen peroxide (SOD). No pattern: contains oxygen and superoxide (catalase). Solid grey: contains oxygen, superoxide, and hydrogen peroxide (standard aerobic reactions). AnO2, anaerobic reactions.

Scheme 1.

Proposed mode of binding of the nitroolefin substrate ( Z )-PNE by PETN reductase. Y-OH refers to the proton donor; either Tyr186 or H 2 O. CG2, methyl group of T26 purported to potentially clash with the binding of some substrates. Adapted from Fryszkowska et al. [20].

The fact that all anaerobic reactions generated predominantly (S)-PNA (Fig. 2) suggests that the presence of oxygen (no ROS) dramatically switches the preference to (R)-PNA production. Conversely, the presence of ROS leads to an increase in the levels of (S)-PNA, with the majority of the effect being seen with superoxide. Therefore, under aerobic conditions, there is competition between oxygen and ROS to increase the levels of (R)-PNA and (S)-PNA, respectively, which leads to an overall poor and/or variable product ee. These effects are buffer-dependent and NAD(P)H-dependent, with the presence of Tris buffer typically leading to a lowering of (S)-PNA ee. Reactions in the presence of both Tris buffer and NADH were more likely to favour (R)-PNA formation than were other reactions.

Scheme 2.

Proposed mechanisms of reaction of oxygen with the FMN cofactor of OYEs to generate superoxide and/or hydrogen peroxide. The initial complex is shown as a caged radical pair of a neutral flavin semiquinone and a superoxide anion. Based on the studies of Massey [26] and Brown et al. [25].

Oxygen influence on product ee

The mechanism of the aerobic, time-dependent switch from the formation of predominantly (R)-PNA to (S)-PNA is unknown, and may be a result of many simultaneously contributory effects. For example, the reaction may be influenced by a variation in the level of dissolved oxygen throughout the reaction. If so, the gradual increase in (S)-PNA production over (R)-PNA production with increasing time may be partly attributable to the reactions becoming progressively anaerobic. Product racemization is unlikely to be affecting the (R)-PNA/(S)-PNA ratio in this case, even in the presence of ROS. Another possible explanation could be that oxygen and/or ROS are promoting the slow nonenzymatic isomerization between (E)-PNE and (Z)-PNE. Prior studies have suggested that substrate binding of (Z)-PNE to PETN reductase is preferential, and the lower ees with (E)-PNE may be attributable to multiple substrate-binding conformations leading to the formation of both enantiomeric products (Scheme 3). A number of models of the binding of both isomers (E/Z) of the nitroolefin PNE to PETN reductase and other OYEs have been proposed, owing to the absence of a substrate-bound crystal structure (Scheme 3) [2, 4, 16]. A shift in the (E)-PNE/(Z)-PNE ratio may affect the proportion of substrate binding in the alternative orientations, leading to a change in product enantiopurity.

Scheme 3.

Proposed mode(s) of binding of the nitroolefin substrates ( Z )-PNE and ( E )-PNE by PETN reductase. Adapted from Fryszkowska et al. [20].

To investigate this, we determined the degree of (E)-PNE to (Z)-PNE isomerization under biotransformation conditions (no PETN reductase), in the presence or absence of SOD and catalase. Superoxide and hydrogen peroxide were generated by the inclusion of both xanthine oxidase (XO) and xanthine [28]. As these reaction conditions often lead to significant substrate decomposition, we evaluated the isomerization after 0 and 6 h. The results showed that the level of isomerization of (E)-PNE to (Z)-PNE was negligible (1–2%) under all biotransformation reaction conditions tested, with only minor variations being observed between anaerobic and aerobic (with and without SOD and catalase) reactions (Table S9). Therefore, substrate isomerization is not responsible for the difference in (R)-PNA/(S)-PNA product ratios between aerobic and anaerobic reactions.

The observed switch in enantiopreference naturally leads to the question of the underlying mechanism(s) for this change in product outcome. A simple, but perhaps unlikely, potential mechanism implicates the involvement of oxygen and ROS ‘binding’ in the active site of PETN reductase. These are probably low-affinity sites, and the binding of an oxygen and/or ROS could influence the orientation of binding of (E)-PNE, thereby affecting the stereochemical outcome of the reaction. However, a more likely mechanism could be the chemical modification of cysteines and/or methionines of PETN reductase by oxygen and/or ROS, with concomitant effects on substrate binding in the active site. However, residues of this type are not found in the active site of PETN reductase, although buried methionines are located nearby (Met22 and Met270) [29]. Additionally, PETN reductase contains one surface-exposed cysteine (Cys222) located on the opposite side of the protein from the active site (Fig. S1).

We subjected PETN reductase to a treatment designed to mimic the oxidative environment present during 24-h biotransformations. This was followed by accurate mass determination by ESI MS to look for the presence of covalent modification of the enzyme. This involved incubating the enzyme aerobically for 24 h in the presence of NADH and in the absence of any additional oxidative substrate. This means that protein turnover can occur with oxygen as the oxidative substrate, generating hydrogen peroxide and superoxide (Scheme 2). Additional control treatments were performed in which PETN reductase was incubated aerobically for 24 h in the absence of NADH, and anaerobically with or without NADH. Analysis by ESI MS gave a mass of 40 455 Da for untreated PETN reductase (Fig. S2A–E), consistent with the presence of a C-terminal His8-tag, and the absence of both the N-terminal methionine and the non-covalently bound FMN. The absence of the N-terminal methionine is consistent with crystallographic studies [29], and is a common form of post-translational modification [30]. Additional minor species in the untreated and NADH-free samples included mass increases consistent with the addition of two and eight oxygen atoms (Fig. S2A,B,D). This suggests that the untreated enzyme had already undergone some minor oxidative modification, as the purification procedures were performed in the absence of any reductant (Fig. S2E).

Interestingly, PETN reductase exposed to oxidative treatment in the presence of NADH showed a significantly altered mass profile, with an additional species present consistent with the addition of six oxygen atoms. Additionally, this suggests that the ratio of unmodified to modified species had decreased significantly, although this is not quantifiable. There was a minor change in the detection of modified species in the anaerobic/NADH samples to above that with NADH-free enzyme, consistent with the presence of a small quantity of oxygen over the 24-h period (< 5 p.p.m). Therefore, under aerobic conditions, PETN reductase is chemically modified to form a mixture of species, with the degree of modification being increased in the presence of ROS. These modifications are likely to include oxidatively modified methonines (e.g. methionine sulfoxide) and cysteines (e.g. sulfinic and sulfonic acids) [31, 32]. These species have not been detected in previous crystal structures of PETN reductase [23, 29], as a minimum of 30% of the protein needs to contain the modification for it to be visible in the electron density.

Because of the likelihood of multiple degrees and sites of modification of PETN reductase, including nontreated samples, it will probably be difficult to prove which, or indeed if any, of the modifications are responsible for the differences in (S/R)-PNA product ee. Therefore, although we have determined that PETN reductase undergoes chemical modification under oxidative conditions, we have not excluded other mechanism(s) that may additionally contribute to (S/R)-PNA enantiopurity changes.

Previous site-directed mutagenesis studies of PETN reductase have shown that subtle changes in the active site can dramatically influence the enantiomeric outcome of (E)-PNE reduction [20]. For example, a mutation of Thr26 to serine, a residue that interacts with the isoalloxazine ring of FMN, results in a switch in product enantiopurity to the formation of predominantly (R)-PNA (37% ee) under anaerobic conditions. Therefore, subtle changes in the active site of PETN reductase can sometimes have a significant effect on the preferred substrate-binding conformation(s), leading to a dramatic change in the product enantiopurity [20]. Given that the exposed cysteines and methionines are distal to the active site, we are probably seeing long-distance effects such as a change in protein dynamics and/or flexibility [33]. The effects of distal mutations on catalytic functioning are not predictive, but can result in changes in catalytic rate, substrate specificity, cofactor dependence, and thermal stability [33-35]. Further mutations to remove the exposed cysteine/methionines may help to increase (S)-PNA enantiopurity, and thereby reduce the oxygen sensitivity of the reaction.

PETN reductase-catalysed reductions of (Z)-PNE are not influenced by the presence of oxygen (Fig. 1), so these oxidative modifications do not affect the range of binding conformations of all substrates. This study highlights the importance of performing reactions of redox enzymes under anaerobic conditions, to minimize any nonproductive side reactions leading to a lowering of the product yield and oxidative damage to the protein. Reaction optimization studies, in which conditions such as reaction time and oxygen presence are varied, are useful to determine whether there is any impact on the reaction yields and/or product enantiopurity. This is especially important for reactions with engineered enzymes, where the full implications of the mutation(s) may not be evident from comparative studies under standard conditions.

Concluding remarks

Our work has revealed unusual opposing effects of oxygen and ROS on the product enantiopreference outcome of PETN reductase-catalysed reactions with the activated alkene substrate (E)-PNE. It emphasizes a need to investigate likely mechanisms for oxygen-dependent and ROS-dependent switches in product enantiopreference. This highlights the importance of carefully defining and controlling reaction conditions in comparative biotransformation reactions with PETN reductase (and potentially other flavoprotein enzymes). Given the current widespread interest in the use of OYEs, these studies are relevant to applications of flavoproteins in synthetic biology and biocatalysis programmes.

Experimental procedures


All reagents were of analytical grade. Anaerobic kinetics and biotransformation reactions were set up and/or monitored within an anaerobic glove box (Belle Technology, Weymouth, UK) under a nitrogen atmosphere (< 5 p.p.m. oxygen). Prior to anaerobic reactions, wild-type and His8-tagged PETN reductase were deoxygenated by passage through a BioRad 10DG column equilibrated in anaerobic reaction buffer (Hemel Hempstead, UK). Aerobic reactions were set up in normal atmospheric concentrations of oxygen. The concentration of substrates was determined by the extinction coefficient method, with previously described values [14, 16, 19]. The alkenes (E/Z)-PNE and ketoisophorone were dissolved as stock solutions in ethanol or dimethylformamide, and then diluted to 5 mm in the reactions (5% final solvent concentration in the reactions). The buffers used in the reactions were phosphate (50 mm KH2PO4/K2HPO4, pH 7.0) or Tris (50 mm Tris, pH 7.5). All medium components were obtained from Formedium (Hunstanton, UK). The proteins SOD, catalase, cytochrome c, HRP and XO were obtained from Sigma-Aldrich (Gillingham, UK).

Chemistry and analytical procedures

All chemicals were obtained from commercial sources, and the solvents were of analytical grade. The nitroalkenes (E)-PNE and (Z)-PNE and their respective nitroalkane products were synthesized as described previously [16, 19]. HPLC analysis was performed with an instrument equipped with a UV detector. Reaction progress was monitored by TLC on standard silica gel plates. The determination of the percentage yield, percentage conversion and percentage ee for all compounds were performed as described previously [16, 19].

Enzyme production and purification

Wild-type nontagged and C-terminally His8-tagged PETN reductase enzymes were prepared as described previously [16, 22]. Purity was assessed by SDS/PAGE (> 90%), and the concentration of active enzyme was determined with the extinction coefficient method by use of a Cary UV-50 Bio UV–visible scanning spectrophotometer and a 1-mL quartz cuvette (Hellma, Southend-on-Sea, UK) with a 1-cm path length (ε464 nm = 11 300 m−1·cm−1) [14, 16].

Standard kinetic analysis of PETN reductase

Standard reactions (1.0 mL) were performed in buffer (phosphate or Tris), PETN reductase (0.015–0.5 μm) and NAD(P)H (100 μm) in the presence or absence of ketoisophorone (100 μm). NAD(P)H oxidation was monitored continuously at 340 nm for 1–2 min at 25 °C. Reactions were performed in the presence or absence of oxygen. Reactions were also performed in the presence of SOD (20 U) and/or catalase (20 U) to determine the effect of the elimination of superoxide and hydrogen peroxide, respectively, on the reaction.

Identification of the ROS generated during biocatalysis

Detection of superoxide production

Standard aerobic reactions (1.0 mL) were performed in buffer (phosphate or Tris), NAD(P)H (100 μm), PETN reductase (0.015–0.5 μm) and cytochrome c (20 μm). The generation of superoxide was detected by monitoring the reduction of cytochrome c at 550 nm (ε550 nm = 21 000 m−1·cm−1) [36].

Endpoint analysis of superoxide formation

Aerobic reactions (1.0 mL) were performed in buffer (phosphate or Tris), NAD(P)H (20 μm), PETN reductase (0.5 μm) and cytochrome c (20 μm) in the presence or absence of excess SOD (20 U) and/or catalase (20 U). Catalase was added to reactions containing SOD to eliminate the reoxidation of reduced cytochrome c by hydrogen peroxide produced in the presence of SOD and superoxide [25]. Reactions were monitored as above until all of the NAD(P)H was consumed, which shortly thereafter resulted in no further absorbance increase at 550 nm (3–5 min with NADPH and 1 h with NADH). Absolute superoxide production was compared with total NAD(P)H consumption.

Endpoint analysis of hydrogen peroxide formation

Standard aerobic reactions (1.0 mL) were performed in buffer (phosphate or Tris), NAD(P)H (20 μm) and PETN reductase (0.5 μm) in the presence or absence of SOD (20 U) and/or catalase (20 U). Reactions were monitored at 340 nm until NAD(P)H depletion was complete (10 min with NADPH and 1 h with NADH). A reaction sample (750 μL) was added to a DAB/gelatin solution (100 mm Na2HPO4/citrate, pH 4.3, containing 9 mg of DAB and 5 mg·mL−1 gelatin; 200 μL) [27] and HRP (1 U total; 50 μL), and vortexed. Colour development was immediate, and the oxidation of DAB by hydrogen peroxide via HRP was detected at 465 nm (ε465 nm = 3160 m−1·cm−1) [27]. Absolute hydrogen peroxide production was compared with total NAD(P)H consumption.

PETN reductase pretreatments

PETN reductase (106 μm) in 50 mm KH2PO4/K2HPO4 (pH 7.0) was incubated at 20 °C for 24 h in the presence or absence of oxygen and NADH (15 mm). NADH and/or oxygen was removed by anaerobic desalting, as above. NADH-containing samples were desalted a second time, as above, to ensure complete removal of the cofactor.


Protein samples (15 μL) were treated with formic acid (0.1%, 5 μL), and then loaded (5 μL) onto a monolithic trap column. Proteins were washed off the column with an acetonitrile gradient into a Waters LCT classic mass spectrometer, with leuenk to provide a lock mass. The data were then analysed with masslynx maxent 1.

Biotransformations with PETN reductase

Standard biotransformations

Standard reactions (1.0 mL) were performed aerobically or anaerobically in phosphate buffer containing alkene (5 mm), PETN reductase (0.002–2.0 μm), and NAD(P)H (6–10 mm). The reactions were shaken at 30 °C for 0.5–48 h at 130 r.p.m., and terminated by extraction with ethyl acetate (0.9 mL) containing an internal standard. The extracts were dried with MgSO4 and analysed by GC or HPLC to determine the percentage yield, percentage conversion, and ee, as described previously [16, 19].

Effect of pH/buffers

Reactions were performed as described for the standard biotransformations, except that a variety of buffer compositions and final pH values were used. Buffer and pH combinations are described in the table legends.

Effect of ROS eliminators on biotransformations

Standard reactions (1.0 mL) were performed as above [5 mm (E)-PNE, 15 mm NAD(P)H, 2 μm PETN reductase] in the presence of excess SOD (20 U) and/or catalase (20 U).

Biotransformations employing an NADPH recycling system

Reactions (1 mL) were performed as described for the effect of ROS eliminators on biotransformations in the presence of NADP+ (10 μm), glucose (15 mm) and GDH (10 U) instead of excess NADPH.

Stability of SOD and catalase under biotransformation conditions

Additional control reactions were performed in the absence of PETN reductase, and analysed at 0 and 24 h for residual SOD and catalase activity. SOD activity was determined with the cytochrome c reduction assay, with XO as the superoxide generator [28]. These reactions (1.0 mL) were performed in buffer (phosphate) containing xanthine (0.15 mm), XO (3.3 mU), cytochrome c (20 U), and biotransformation reaction sample (50 μL). SOD activity was detected as a decrease in the rate of cytochrome c reduction monitored at 550 nm, as both SOD and cytochrome c compete for the superoxide generated by xanthine/XO [28]. The detection of residual catalase activity was performed by combining phosphate buffer (50 μL) with a hydrogen peroxide/water solution (30%; 50 μL) and a biotransformation reaction sample (50 μL). Activity was detected qualitatively by the visual detection of oxygen bubble formation. No quantitative catalase activity assay (hydrogen peroxide spectral change or DAB/HRP assay) could be performed, owing to spectral interference between reaction components.

Substrate isomerization studies

Reactions (1 mL) were performed aerobically in buffer (phosphate or Tris), (E)-PNE (5 mm), NADP+ (10 μm), glucose (15 mm), GDH (10 U), xanthine (0.15 mm), and XO (3.3 mU), in the presence of absence of SOD (20 U) and/or catalase (20 U). Anaerobic control reactions were performed in anaerobic buffers in the absence of ROS eliminators. All reactions were agitated at 30 °C and 130 r.p.m., and analysed at 0 and 6 h by HPLC with a Chiracel OD column [hexane/i-PrOH, 9 : 1, (Z)-PNE and (E)-PNE, 8.9 and 10.7 min, respectively].


This work was funded by the UK Biotechnology and Biological Sciences Research Council (BBSRC). N. S. Scrutton is a Royal Society Wolfson Merit Award holder and an Engineering and Physical Sciences Research Council (EPSRC) Established Career Fellow. G. Srephens is a BBSRC Research Development Fellow.