University of Groningen Characterization and Crystal Structure of a Robust Cyclohexanone Monooxygenase

Cyclohexanone monooxygenase (CHMO) is a promising biocatalyst for industrial reactions owing to its broad substrate spectrum and excellent regio-, chemo-, and enantioselectivity. However, the low stability of many Baeyer– Villiger monooxygenases is an obstacle for their exploitation in industry. Characterization and crystal structure determination of a robust CHMO from Thermocrispum municipale is reported. The enzyme efficiently converts a variety of aliphatic, aromatic, and cyclic ketones, as well as prochiral sulfides. A compact substrate-binding cavity explains its preference for small rather than bulky substrates. Small-scale conversions with either purified enzyme or whole cells demonstrated the remarkable properties of this newly discovered CHMO. The exceptional solvent tolerance and thermostability make the enzyme very attractive for biotechnology. Cyclohexanone monooxygenase (CHMO; EC 1.14.13.22) is an FADand NADPH-dependent Baeyer–Villiger monooxygenase (BVMO). A wide variety of ketones are converted by CHMO into esters or lactones through the insertion of an oxygen atom on one side or the other of the carbonyl group. In addition, CHMO oxidizes aldehydes and heteroatoms and carries out epoxidation reactions. In contrast to conventional Baeyer–Villiger oxidations using peracids, CHMO reactions are environmentally friendly and often proceed with excellent regio-, chemo-, and enantioselectivity. Numerous industrial applications have been suggested for CHMO. The conversion of cyclohexanone as catalyzed by CHMO is of particular interest since the product, e-caprolactone, is a precursor of both adipic acid and e-caprolactam, which are known polymer building blocks. One of the main barriers to exploiting CHMO as a biocatalyst is its lack of stability. To date, the best-characterized CHMOs are from Acinetobacter calcoaceticus NCIMB 9871 (AcCHMO) and Rhodococcus sp. HI-31 (RhCHMO). Similar CHMOs have been identified from various bacteria (Table S1 in the Supporting Information). However, engineering attempts to improve their stability have met with limited success. The only robust BVMO described so far, phenylacetone monooxygenase from Thermobifida fusca (TfPAMO; EC 1.14.13.92), shows no activity on cyclohexanone. Our interest in using a robust BVMO to produce e-caprolactone and other valuable compounds prompted us to search for a stable CHMO in the available genomes of thermophiles. Among the retrieved sequences, a putative CHMO from Thermocrispum municipale DSM 44069 was found (TmCHMO; NCBI RefSeq: WP_028849141.1). This thermophilic organism was isolated from municipal waste compost. TmCHMO clusters with known CHMOs sequences (Figures S1, S2 in the Supporting Information), the closest homologue being from Brachymonas petroleovorans (66% sequence identity). We identified three genes upstream of the TmCHMO gene that may participate in cyclohexanol degradation (Table S2). These sequence analyses suggested that the T. municipale enzyme may be a robust CHMO. Therefore, we purified TmCHMO and studied its substrate acceptance (Figure S3). For the sake of comparison, the same analysis was performed with AcCHMO. Small aliphatic ketones such as cyclohexanone, cyclobutanone, and bicyclo[3.2.0]hept-2-en-6-one were efficiently converted by both CHMOs, although in all cases, the Km values for AcCHMO were higher than those measured for TmCHMO, despite similar kcat values (Table S3). Conversely, the bulkier molecules progesterone and cyclopentadecanone were not substrates of either enzyme. At high substrate concentrations, substrate inhibition was observed for TmCHMO, which is not unprecedented for this enzyme class (Figure S4). To avoid decreased conversions as a result of this effect, reaction media with two phases have been successfully implemented. The uncoupling rate for both CHMOs was around 0.1 s . To compare their thermostability, the ThermoFAD method was used as a first approach. This confirmed that AcCHMO is only moderately stable since it exhibits a Tm value of 37 8C (Table S4). Conversely, the Tm value for TmCHMO was found to be 11 8C higher. Next, the two CHMOs were probed for their resistance to thermal inactivation (Figure 1A). This revealed that TmCHMO still displayed 58 % residual activity after 5.5 h at 45 8C. By contrast, AcCHMO lost its activity within a few minutes at 45 8C. These experiments were complemented by analyzing the effect of cosolvents on the stability, since enzymes working in organic solvents or aqueous/organic mixtures are often desired for biotechnology. While AcCHMO essentially lost its activity after [*] Dr. E. Romero, Prof. Dr. M. W. Fraaije Department of Biotechnology, University of Groningen Nijenborgh 4, 9747AG, Groningen (The Netherlands) E-mail: m.w.fraaije@rug.nl Dr. J. R. G. Castellanos, Prof. Dr. A. Mattevi Department of Biology and Biotechnology “Lazzaro Spallanzani” University of Pavia, Via Ferrata 9, 27100, Pavia (Italy) E-mail: andrea.mattevi@unipv.it [] These authors contributed equally to this work. Supporting information and the ORCID identification number(s) for the author(s) of this article can be found under http://dx.doi.org/10. 1002/anie.201608951. 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. Angewandte Chemie Communications 1 Angew. Chem. Int. Ed. 2016, 55, 1 – 5 2016 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim These are not the final page numbers! 25 min incubation in 14 % acetonitrile at 20 8C, TmCHMO remained highly active (> 80 %) for at least 20 h under these conditions (Figure 1B). These thermostability and solventtolerance data clearly show that TmCHMO is a substantially more robust biocatalyst than AcCHMO. Besides influencing stability, the reaction medium can also modify enzyme selectivity. The effect of cosolvents (Table S5) was determined by using 2-butanone, the conversion of which to two possible regioisomers is of industrial interest (Scheme S1). All reactions were stopped after 48 h at 17 8C. The ratio between the products methyl propanoate and ethyl acetate was found to be about 3:7 for both purified TmCHMO and AcCHMO in the absence of any cosolvent (Figure S5). The same regioisomer ratio was observed using whole cells of Escherichia coli expressing one or the other CHMO (not shown). Next, we inspected the effect of various solvents at 15 % concentration. The two purified CHMOs exhibited similar results, although the yields of TmCHMO were generally higher. The strongest effect on regioselectivity was observed with 2-methyl-1,3-dioxolane, which led to almost exclusive production of ethyl acetate. 1,3-Dioxane and 1,4-dioxane had a more moderate influence, since about 40% of the total product was methyl propanoate. We also carried out reactions with 30 % methanol or ethanol. These cosolvents had a negligible effect on enzyme regioselectivity. However, they considerably decreased the 2-butanone conversion yield for AcCHMO (< 4%), while that for TmCHMO remained high (96% and 56% for methanol and ethanol, respectively). From these results, it can be concluded that the robustness of TmCHMO makes possible to modulate its regioselectivity through using cosolvents. Along the lines of the previous experiments, the potential of TmCHMO as a regioselective biocatalyst was further probed by using rac-bicyclo[3.2.0]hept-2-en-6-one. This compound can be converted by BVMOs into four products (Scheme S2). This reaction is widely used to study the ability of BVMOs to carry out the kinetic resolution of racemic compounds, and it is of interest for the synthesis of prostaglandins, for example. Small-scale conversions of rac-bicyclo[3.2.0]hept-2-en-6-one were carried out with TmCHMO, AcCHMO, and TfPAMO. Both enantiomers of this ketone were fully converted by both CHMOs, yielding almost exclusively one regioisomer from each enantiomer (Figure S6A and Table S6). By contrast, TfPAMO produced all four possible lactones, proving to be far less regioselective than CHMOs. We also used the same BVMOs to produce enantiomerically pure sulfoxides, which are widely used in asymmetric synthesis and are often biologically active. The prochiral compound thioanisole was chosen as a model substrate (Scheme S3). The CHMOs exclusively produced the (R)-sulfoxide, whereas TfPAMO produced both enantiomers, leading to an ee of only 16 % for the (R)-sulfoxide (Figure S6B, Table S6). Having established that TmCHMO is an appealing biocatalyst based on its thermostability, solvent tolerance, and selectivity, we performed a more in-depth characterization of its mechanistic properties. The reaction mechanism of a BVMO generally involves a C4a-peroxyflavin intermediate that forms a tetrahedral Criegee intermediate through nucleophilic attack on the substrate carbonyl carbon (Scheme S4). Rearrangement of the Criegee intermediate yields the ester or lactone product. The spectral changes for TmCHMO during its catalytic cycle were monitored using a stopped-flow spectrophotometer. Anaerobic reaction of TmCHMO with NADPH resulted in the loss of the absorbance peaks at 376 nm and 440 nm, which is consistent with the formation of the two-electron-reduced enzyme. After mixing the reduced TmCHMO with air-saturated buffer, a rapid increase in absorbance at 355 nm was observed (k = 37 s ), together with a small absorbance decrease at 450 nm (Figure 2A). These spectral changes are indicative of the formation of the C4a-peroxyflavin intermediate. The absorbance at 355 nm was stable for 3 s and then slowly decreased (k = 0.01 s ) owing to decay of the intermediate, which is consistent with hydrogen peroxide elimination to form the reoxidized enzyme (k = 0.004 s ). In a second set of experiments, the anaerobically reduced TmCHMO was mixed with cyclohexanone in air-saturate

Cyclohexanone monooxygenase (CHMO;EC1.14.13.22) is an FAD-and NADPH-dependent Baeyer-Villiger monooxygenase (BVMO). [1] Awide variety of ketones are converted by CHMO into esters or lactones through the insertion of an oxygen atom on one side or the other of the carbonyl group. In addition, CHMO oxidizes aldehydes and heteroatoms [2] and carries out epoxidation reactions. [3] In contrast to conventional Baeyer-Villiger oxidations using peracids,C HMO reactions are environmentally friendly and often proceed with excellent regio-, chemo-, and enantioselectivity. [4] Numerous industrial applications have been suggested for CHMO. [5] Thec onversion of cyclohexanone as catalyzed by CHMO is of particular interest since the product, e-caprolactone,isaprecursor of both adipic acid and e-caprolactam, [6] which are known polymer building blocks. [7] One of the main barriers to exploiting CHMO as ab iocatalyst is its lack of stability. [8] To date,t he best-characterized CHMOs are from Acinetobacter calcoaceticus NCIMB 9871 (AcCHMO) and Rhodococcus sp.H I-31 (RhCHMO). Similar CHMOs have been identified from various bacteria ( Table S1 in the  Supporting Information). However,e ngineering attempts to improve their stability have met with limited success. [9] The only robust BVMO described so far,p henylacetone monooxygenase from Thermobifida fusca (TfPAMO;E C 1.14.13.92), shows no activity on cyclohexanone. [10] Our interest in using ar obust BVMO to produce e-caprolactone and other valuable compounds prompted us to search for as table CHMO in the available genomes of thermophiles. Among the retrieved sequences,aputative CHMO from Thermocrispum municipale DSM 44069 was found (TmCHMO;NCBI RefSeq:WP_028849141.1). This thermophilic organism was isolated from municipal waste compost. [11] TmCHMO clusters with known CHMOs sequences (Figures S1, S2 in the Supporting Information), the closest homologue being from Brachymonas petroleovorans (66 % sequence identity). We identified three genes upstream of the TmCHMO gene that may participate in cyclohexanol degradation (Table S2). These sequence analyses suggested that the T. municipale enzyme may be ar obust CHMO.
Therefore,w ep urified TmCHMO and studied its substrate acceptance ( Figure S3). [4,12] Forthe sake of comparison, the same analysis was performed with AcCHMO.S mall aliphatic ketones such as cyclohexanone,c yclobutanone,and bicyclo[3.2.0]hept-2-en-6-one were efficiently converted by both CHMOs,a lthough in all cases,t he K m values for AcCHMO were higher than those measured for TmCHMO, despite similar k cat values (Table S3). Conversely,the bulkier molecules progesterone and cyclopentadecanone were not substrates of either enzyme. [13] At high substrate concentrations,substrate inhibition was observed for TmCHMO,which is not unprecedented for this enzyme class ( Figure S4). To avoid decreased conversions as aresult of this effect, reaction media with two phases have been successfully implemented. [5] Theuncoupling rate for both CHMOs was around 0.1 s À1 .T o compare their thermostability,t he ThermoFAD method was used as af irst approach. [14] This confirmed that AcCHMO is only moderately stable since it exhibits a T m value of 37 8 8C (Table S4). [9a,b] Conversely,t he T m value for TmCHMO was found to be 11 8 8Chigher. Next, the two CHMOs were probed for their resistance to thermal inactivation ( Figure 1A). This revealed that TmCHMO still displayed 58 %residual activity after 5.5 ha t4 58 8C. By contrast, AcCHMO lost its activity within af ew minutes at 45 8 8C. These experiments were complemented by analyzing the effect of cosolvents on the stability,s ince enzymes working in organic solvents or aqueous/organic mixtures are often desired for biotechnology. [15] While AcCHMO essentially lost its activity after 25 min incubation in 14 %a cetonitrile at 20 8 8C, TmCHMO remained highly active (> 80 %) for at least 20 hunder these conditions ( Figure 1B). These thermostability and solventtolerance data clearly show that TmCHMO is asubstantially more robust biocatalyst than AcCHMO.
Besides influencing stability,the reaction medium can also modify enzyme selectivity. [15][16] Thee ffect of cosolvents (Table S5) was determined by using 2-butanone,t he conversion of which to two possible regioisomers is of industrial interest (Scheme S1). [17] All reactions were stopped after 48 h at 17 8 8C. Ther atio between the products methyl propanoate and ethyl acetate was found to be about 3:7f or both purified TmCHMO and AcCHMO in the absence of any cosolvent ( Figure S5). Thes ame regioisomer ratio was observed using whole cells of Escherichia coli expressing one or the other CHMO (not shown). Next, we inspected the effect of various solvents at 15 %c oncentration. Thet wo purified CHMOs exhibited similar results,a lthough the yields of TmCHMO were generally higher. Thestrongest effect on regioselectivity was observed with 2-methyl-1,3-dioxolane,w hich led to almost exclusive production of ethyl acetate.1 ,3-Dioxane and 1,4-dioxane had am ore moderate influence,s ince about 40 %o ft he total product was methyl propanoate.W ea lso carried out reactions with 30 %m ethanol or ethanol. These cosolvents had anegligible effect on enzyme regioselectivity. However,t hey considerably decreased the 2-butanone conversion yield for AcCHMO (< 4%), while that for TmCHMO remained high (96 %a nd 56 %f or methanol and ethanol, respectively). From these results,itcan be concluded that the robustness of TmCHMO makes possible to modulate its regioselectivity through using cosolvents.
Along the lines of the previous experiments,the potential of TmCHMO as ar egioselective biocatalyst was further probed by using rac-bicyclo[3.2.0]hept-2-en-6-one.T his compound can be converted by BVMOs into four products (Scheme S2). This reaction is widely used to study the ability of BVMOs to carry out the kinetic resolution of racemic compounds, [18] and it is of interest for the synthesis of prostaglandins,f or example. [19] Small-scale conversions of rac-bicyclo[3.2.0]hept-2-en-6-one were carried out with TmCHMO,A cCHMO,a nd TfPAMO.B oth enantiomers of this ketone were fully converted by both CHMOs,y ielding almost exclusively one regioisomer from each enantiomer ( Figure S6A and Table S6). [20] By contrast, TfPAMO produced all four possible lactones,p roving to be far less regioselective than CHMOs.W ea lso used the same BVMOs to produce enantiomerically pure sulfoxides,w hich are widely used in asymmetric synthesis and are often biologically active. [21] Thep rochiral compound thioanisole was chosen as am odel substrate (Scheme S3). TheC HMOs exclusively produced the (R)-sulfoxide,w hereas TfPAMO produced both enantiomers,leading to an ee of only 16 %for the (R)-sulfoxide ( Figure S6B,T able S6). [18c,22] Having established that TmCHMO is an appealing biocatalyst based on its thermostability,s olvent tolerance, and selectivity,w ep erformed am ore in-depth characterization of its mechanistic properties.The reaction mechanism of aBVMO generally involves aC4a-peroxyflavin intermediate that forms at etrahedral Criegee intermediate through nucleophilic attack on the substrate carbonyl carbon (Scheme S4). Rearrangement of the Criegee intermediate yields the ester or lactone product. [23] Thespectral changes for TmCHMO during its catalytic cycle were monitored using as topped-flow spectrophotometer.A naerobic reaction of TmCHMO with NADPH resulted in the loss of the absorbance peaks at 376 nm and 440 nm, which is consistent with the formation of the two-electron-reduced enzyme.After mixing the reduced TmCHMO with air-saturated buffer, ar apid increase in absorbance at 355 nm was observed (k = 37 s À1 ), together with as mall absorbance decrease at 450 nm (Figure 2A). These spectral changes are indicative of the formation of the C4a-peroxyflavin intermediate.T he absorbance at 355 nm was stable for 3sand then slowly decreased (k = 0.01 s À1 )o wing to decay of the intermediate,w hich is consistent with hydrogen peroxide elimination to form the reoxidized enzyme (k = 0.004 s À1 ). In as econd set of experiments,the anaerobically reduced TmCHMO was mixed with cyclohexanone in air-saturated buffer.T he absorbance at 355 nm increased for 0.1 sa nd then immediately decreased, which demonstrates the low kinetic stability of the intermediate in the presence of cyclohexanone ( Figure 2B and Figure S7). Therate of formation of the peroxyflavin was not influenced by the presence of cyclohexanone,while its decay rate was 80-fold higher than that measured in the absence of this ketone.C ollectively,t hese experiments suggest that TmCHMO functions as at ypical BVMO,f orming as table flavin peroxide that can effectively perform substrate oxygenation. Thekinetic stability of the peroxyflavin enables the enzyme to efficiently couple NADPH and dioxygen consumption with substrate oxygenation without leakage of hydrogen peroxide,w hich can be harmful in the context of large-scale biotransformations. [23a-c] Foramore in-depth understanding of TmCHMO properties,its crystal structure in complex with FADand NADP + in the oxidized and reduced states were solved to aresolution of 1.22 and 1.60 (Table S7;F igure 3A,a nd Figure S8). In an attempt to rationalize the relatively high thermostability,w e investigated the number of salt bridges present, since they are known to contribute to the thermostability of proteins. [24] Our computational analysis of TfPAMO,T mCHMO,a nd RhCHMO identified 37, 31, and 16 salt bridges,r espectively ( Figure S9). This finding correlates with their T m values (61, 48, and 37 8 8C, respectively). Mutagenesis on these BVMOs will be carried out to confirm the role of salt bridges in their stability.
Theo verall structures of the oxidized and reduced forms share an almost identical conformation, with ar oot mean square deviation of 0.20 for the backbone Ca atoms. However,i nspection of the active sites shows distinct alterations ( Figure 3B and Figure S10). In the reduced enzyme, the electron density corresponding to the nicotinamide moiety of NADP + is disordered, with no well-defined electron density.F urthermore,R 329 of the oxidized enzyme is engaged in H-bonds with the carboxamide group of NADP + and the side chain of D59. Upon enzyme reduction, R329 moves away from NADP + and points toward the isoallox-azine moiety of the flavin ring, which favors an electrostatic interaction between the positively charged guanidinium group of R329 and the negatively charged reduced flavin. Upon formation of the flavin peroxide and concomitant loss of the negative charge on the flavin ring, R329 would shift back to the conformation interacting with the NADP + , thereby making the catalytic center accessible to ak etone substrate. [23d] These results confirm that the formation of the negatively charged reduced flavin is associated with al ocalized rearrangement of the central elements of the catalytic site.N ADP + is required for peroxyflavin formation and stabilization, primarily to provide essential H-bonding interactions.
Inspecting the electron density of the oxidized structure of TmCHMO revealed that the putative substrate-binding pocket is occupied by ar ing-shaped ligand ( Figure 4A and Figure S11). This region of electron density was putatively assigned to am olecule of nicotinamide,p ossibly resulting from the degradation of NADP + .S pecifically,t he ligand is placed right in front of the flavin ring and is in contact with residues T60, L145, L428, P430, F434, T435, and L437. This arrangement is very similar to that observed in the structures of RhCHMO tight [23e] and TfPAMO MES , [23d] both with ab ound ligand ( Figure 4B and Figure S12). Theligand-binding site is ac ompact cavity,w hich explains the general preference of TmCHMO for medium/small substrates.
To conclude,w er eport the discovery of ar obust CHMO that shows great promise as an oxidative biocatalyst. The enzyme was found to be much more thermostable and solvent tolerant than known CHMOs.F urthermore,h aving established an effective recombinant production system and elucidated its crystal structure,T mCHMO provides the perfect starting point for engineering approaches to tune its properties.   Programme Research and Innovation actions H2020-LEIT BIO-2014-1 and by the MEBIO grant (053.24.105) from the NWO. We acknowledge the European Synchrotron Radiation Facility and the Swiss Light Source for the provision of beam time.W eappreciate the assistance with the microspectrophotometry measurements of Dr. Antoine Royant of the European Synchrotron Radiation Facility and the "Institut de Biologie Structurale".