H2O2

H2O

DOCTORAL THESIS Carmen Aranda Oliden

BIOTRANSFORMATIONS OF INDUSTRIAL INTEREST CATALYZED BY FUNGAL PEROXYGENASES

Thesis submitted by:

María del Carmen Aranda Oliden

for the fulfillment of the requirements for the degree of doctor (Ph.D.) from the University of Seville.

Biotransformations of industrial interest catalyzed by fungal peroxygenases

Sevilla, 5 de abril de 2019

LOS DIRECTORES

Dra. Dña. Ana Gutiérrez Suárez Dr. D. José C. del Río Andrade

Profesor de Investigación del CSIC Profesor de Investigación del CSIC IRNAS-CSIC IRNAS-CSIC

EL TUTOR

Dr. D. José María Fernández-Bolaños

Catedrático de la Universidad de Sevilla

Memoria que presenta:

María del Carmen Aranda Oliden

Para optar al grado de Doctor por la Universidad de Sevilla.

DOCTOR D. JOSÉ ENRIQUE FERNÁNDEZ LUQUE, DIRECTOR DEL INSTITUTO DE RECURSOS NATURALES Y AGROBIOLOGÍA DE SEVILLA DEL CONSEJO SUPERIOR DE INVESTIGACIONES CIENTÍFICAS.

CERTIFICA: Que la presente Memoria de Investigación titulada “Biotransformations of industrial interest catalyzed by fungal peroxygenases”, presentada por María del Carmen Aranda Oliden para optar al grado de Doctor, ha sido realizada en el Instituto de Recursos Naturales y Agrobiología de Sevilla (IRNAS) del Consejo Superior de Investigaciones Científicas (CSIC), bajo la dirección de los Dres. Dña. Ana Gutiérrez Suárez y D. José Carlos del Río Andrade, reuniendo todas las condiciones exigidas a los trabajos de Tesis Doctorales.

En Sevilla, a 18 de marzo de 2019

AGRADECIMIENTOS

Esta Tesis se ha llevado a cabo en el Instituto de Recursos Naturales y Agrobiología de Sevilla (IRNAS) del Consejo Superior de Investigaciones Científicas (CSIC) con un contrato predoctoral del Ministerio de Economía y Competitividad asociado al Proyecto Nacional “Modificación enzimática de lignina y lípidos en las biorrefinerías de la lignocelulosa” (BIORENZYMERY, AGL2014-53730-R, cofinanciado con fondos FEDER). Esta Tesis ha sido financiada también por los Proyectos Europeos “Industrial ” (INDOX, GA-KBBE-2013-7- 613549) y “New enzymatic oxidation /oxyfunctionalization technologies for added value bio-based products” (EnzOx2, H2020-BBI-PPP-2015-2-1-720297).

Mi más sincero agradecimiento:

A los Prof. Dres. Ana Gutiérrez y José Carlos de Río, directores de esta Tesis, por haber creado un grupo tan maravilloso y haberme dado la oportunidad de formar parte de él. Gracias por todo el trabajo tan duro que lleváis a cabo día tras día, por formarme profesionalmente con vuestras conversaciones a nivel científico y personal, y por todo el apoyo y ayuda necesarios para la realización de esta Tesis.

Al Prof. Dr. Ángel T. Martínez, del Centro de Investigaciones Biológicas (CIB- CSIC, Madrid), por el esfuerzo dedicado a los proyectos europeos y a los trabajos publicados durante esta Tesis. Me gustaría darle las gracias también por su paciencia y tiempo dedicado a las numerosas discusiones científicas que hemos tenido, que tanto conocimiento y motivación me aportaron.

Al Prof. Dr. José María Fernández-Bolaños, Catedrático de la Universidad de Sevilla, tutor de esta Tesis, por su dedicación y ayuda tanto en la parte burocrática como científica.

A los Dres. Katrin Scheibner y Jan Kiebist, de la empresa JenaBios (Jena, Alemania), por suministrar las peroxigenasas de Marasmius rotula y Chaetomium globosum.

Al Prof. Dr. Martin Hofrichter y al Dr. René Ullrich (Universidad Técnica de Dresde, Alemania) por suministrar la peroxigenasa de Agrocybe aegerita.

A la empresa Novozymes A/S (Bagsvaerd, Dinamarca) por suministrar las peroxigenasas recombinantes de Coprinopsis cinerea y Humicola insolens.

Al Dr. Víctor Guallar (Centro Nacional de Supercomputación de Barcelona), por la realización de los estudios computacionales presentados en esta Tesis.

Al Dr. Frank Hollmann (Universidad Técnica de Delft, Países Bajos), por la oportunidad de trabajar con él durante mi estancia, y a todo su grupo BOC por hacer de mi tiempo allí una de las mejores experiencias de mi vida.

A mi inmejorable grupo del IRNAS, por ayudarme con su locura a que mi día a día fuese más divertido y a que los malos momentos supieran menos amargos. Muchas gracias por vuestra paciencia, que no os ha faltado ni un momento, aunque yo os lo haya puesto difícil. Muchas gracias por vuestro apoyo y por hacer que el desarrollo de esta Tesis fuese más fácil.

A mi familia y amigos, en especial a mis padres y abuela, por hacerme la persona que soy hoy en día, por ayudarme todos los días para que yo pudiese realizar esta Tesis, y por todo el interés que mostráis hacia mi trabajo.

ABBREVIATIONS

4HIP 4-hydroxyisophorone 4KIP 4-ketoisophorone 7HIP 7-hydroxyisophorone 7FIP 7-formylisophorone ԑ Molar Extinction Coefficient AaeUPO Unspecific Peroxygenase from Agrocybe aegerita AAO Aryl-alcohol oxidase ABTS 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid) APO Aromatic Peroxygenase CglUPO Unspecific Peroxygenase from Chaetomium globosum CraUPO Unspecific Peroxygenase from Coprinellus radians DHS 4,4′-dihydroxy-trans-stilbene E Epoxide ee Enantiomeric excess EC Commission ED Epoxidized Derivative FA Fatty Acid FAME Fatty Acid Methyl Ester FDCA Furandicarboxylic acid FFCA 2,5-Formylfurancarboxylic acid GC-MS Gas Chromatography – Mass Spectrometry HD Hydroxylated derivative HMF 5-Hydroxymethylfurfural HPLC High Performance Liquid Chromatography IP α-Isophorone IUBMB International Union of Biochemistry and Molecular Biology kcat Catalytic constant

Km Michaelis-Menten constant kcat/ Km Catalytic efficiency MroUPO Unspecific Peroxygenase from Marasmius rotula NADPH Nicotinamide Adenine Dinucleotide Phosphate OxyPin Oxypinosylvin OxyRSV Oxyresveratrol P450 monooxygenase PELE Protein Energy Landscape Exploration Pin Pinosylvin rCciUPO Unspecific Peroxygenase from Coprinopsis cinerea rHinUPO Unspecific Peroxygenase from Humicola insolens RSV Resveratrol St trans-Stilbene St-epoxide trans-Stilbene epoxide TTN Total Turnover Numbers TOF Turnover frequency UPO Unspecific Peroxygenase

INDEX

ABSTRACT……………………………………………………………... 1 1. INTRODUCTION ………………………………………………. 7 1.1 Biocatalytic oxyfunctionalization reactions ………..…….…... 7 1.2 Oxyfunctionalization biocatalysts: Fungal peroxygenases….... 8 1.3 Reaction cycle and mechanism of peroxygenases….………… 12 1.4 Peroxygenase-catalyzed reactions….…………………………. 14 1.4.1 Oxygenation of C-H bonds…………………….………… 15 1.4.2 Oxygenation of C=C bonds .…………………………….. 18 1.4.3 Oxygenation of alcohols and aldehydes………………….. 19 1.4.4 Oxygenation of heteroatoms …………………………….. 21 2. OBJECTIVES …………………………………………………… 25 3. GENERAL RESULTS AND DISCUSSION……...……………. 29 3.1 Oxygenation of trans-stilbene and other stilbenoids by UPOs 29 3.1.1 Oxygenation of trans-stilbene…………………………... 30 Regioselectivity in trans-stilbene oxygenation ………..… 30 Kinetic studies of trans-stilbene oxygenation...………..... 31 3.1.2 Oxygenation of other stilbenoids………………………… 32 3.1.3 Structure-activity analysis….…………………………….. 33 3.2 Oxygenation of α-isophorone by UPOs………………………. 34 3.2.1 Regioselectivity in α-isophorone hydroxylation ……….... 35 3.2.2 Kinetic studies of α-isophorone hydroxylation….……….. 38 3.2.3 Computational analysis…………………………………... 39 3.3 Epoxidation of unsaturated fatty acids and methyl esters by UPOs….……….……………………………………..……….. 40 3.3.1 Regioselectivity of the oxygenation of unsaturated fatty 40 acids and methyl esters…………………………..……….. 3.3.2 Kinetic studies of fatty acids oxygenation ……………… 43 4. REFERENCES…..………………………………………………. 54

5. PUBLICATIONS ..………………………………………………. 59 5.1 Scientific articles…..………………………………………….. 59 Publication I: Aranda C., Ullrich R., Kiebist J., Scheibner K., del Río J.C., Hofrichter M., Martínez A.T. and Gutiérrez A. (2018) Selective synthesis of the resveratrol analogue 4,4′- dihydroxy-trans-stilbene and stilbenoids modification by fungal peroxygenases. Catalysis Science & Technology 8: 2394-2401……………………………………………………... 61 Publication II: Aranda C., Municoy M., Guallar V., Kiebist J., Scheibner K., Ullrich R., del Río J.C., Hofrichter M., Martínez A.T. and Gutiérrez A. (2019) Selective synthesis of 4-hydroxyisophorone and 4-ketoisophorone by fungal peroxygenases. Catalysis Science and Technology (Published online). DOI: 10.1039/C8CY02114G.………………………… 89 Publication III: Aranda C., Olmedo A., Kiebist J., Scheibner K., del Río J.C., Martínez A.T. and Gutiérrez A. (2018) Selective epoxidation of fatty acid and fatty acids methyl esters by fungal peroxygenases. ChemCatChem 10: 3964- 3968………………...... 121 5.2 Patents…………………………..…………………………….. 152 Patent I: Aranda Oliden C., del Río Andrade J. C., Martínez Ferrer A. T., Gutiérrez Suárez A. (2018) Process for the selective synthesis of 4-hydroxyisophorone and 4- ketoisophorone by fungal peroxygenases. EP 18382872 ...... 152 Patent II: Fernández Fueyo E., Aranda Oliden C., Gutiérrez Suárez A., Martínez Ferrer A. T. (2018) Method of heterologous expression of active fungal unspecific peroxygenase in bacterial host cells for fatty-acid epoxidation and other oxygenation reactions. EP 18382514 ………………. 158 6. CONCLUSIONS ...………………………………………………. 171 7. ANNEX…………………...…………..…………………………... 177 7.1 Publication: Olmedo A., Aranda C., del Río J.C., Kiebist J., Scheibner K., Martínez A.T. and Gutiérrez A. (2016) From alkanes to carboxylic acids: Terminal oxygenation by a fungal peroxygenase. Angewandte Chemie International Edition 55:12248-12251 ………………………….……………....…… 177

Abstract

In this Thesis, the selective synthesis of different products of industrial interest such as antioxidants derived from trans-stilbene and other stilbenoids, flavour-and- fragrance additives derived from α-isophorone, and epoxides from unsaturated fatty acids (FA) and their methyl esters (FAME) catalyzed by fungal unspecific peroxygenases (UPOs, EC.1.11.2.1) was studied. With this purpose, several UPOs were tested, including those from the basidiomycetes Agrocybe aegerita (AaeUPO), Marasmius rotula (MroUPO) as well as the recombinant enzyme from Coprinopsis cinerea (rCciUPO) and those from the ascomycetes Chaetomium globosum (CglUPO) and the recombinant enzyme from Humicola insolens (rHinUPO).

The first reaction studied was the oxygenation of trans-stilbene and other stilbenoids such as pinosylvin (Pin) and resveratrol (RSV). In trans-stilbene oxygenation, different reactivities were found depending on the UPO used. AaeUPO, MroUPO and rCciUPO selectively hydroxylated the aromatic rings to form the compound of interest 4,4′-dihydroxy-trans-stilbene (DHS). DHS is a RSV analogue whose preventive effects on cancer invasion and metastasis has recently been shown. The kinetic studies revealed AaeUPO as the most efficient enzyme with one and two orders of magnitude higher catalytic efficiencies (kcat/Km) for the first and second hydroxylation steps, respectively, compared to MroUPO and rCciUPO. Likewise, AaeUPO achieved the highest total turnover numbers (TTN) of up to 200 000. In addition, AaeUPO gave also the best results in Pin and RSV hydroxylation reactions with higher regioselectivity and substrate conversions. However, CglUPO failed to hydroxylate the aromatic ring and instead epoxidized the double bond of the alkenyl moiety forming trans-stilbene epoxide as the only product. True peroxygenative activity was demonstrated by incorporation of 18O 18 from H2 O2 in the oxidation products.

The second reaction studied comprises the α-isophorone hydroxylation by UPOs to produce 4-hydroxyisophorone (4HIP) and 4-ketoisophorone (4KIP), which are flavor-and-fragrances molecules commonly used as additives and as intermediates in the synthesis of vitamins and carotenoids. rHinUPO and CglUPO selectively transformed the substrate (10 mM) in the compounds of interest attaining TTN of up to 5500. Interestingly, although AaeUPO was found to be less selective since this enzyme oxygenated positions 4 and 7 of α-isophorone, it was the only stereoselective enzyme producing the S-enantiomer of 4HIP with an enantiomeric

excess of 88%. On the other hand, using the racemic 4HIP as substrate, a faster conversion of the S-enantiomer by rHinUPO and CglUPO was observed, leading to a kinetic resolution of the racemate with 60-75% recovery of the R-enantiomer. Surprisingly, MroUPO and rCciUPO failed to transform α-isophorone. These differences in regioselectivities could be explained by computational analysis carried out with adaptive PELE software, where differences in the distances between the substrates atoms and the oxo- of reactive Compound I and different binding energies were observed.

The last reaction studied included the oxygenation of FAs and FAMEs to form valuable reactive epoxides. A series of mono- and poly-unsaturated FAs and FAMEs were transformed by MroUPO and CglUPO with high conversion yields, generally attaining better selectivity towards the desired epoxide with the latter enzyme. The differences were observed during oleic acid oxygenation reactions, where CglUPO showed one order of magnitude higher catalytic efficiency and higher TTN (8000) compared to MroUPO (4000). The regioselectivity observed is different from the previously described subterminal hydroxylation by the well- known rCciUPO and AaeUPO, which do not epoxidize unsaturated FAs.

In conclusion, biotransformations using these novel and appealing biocatalysts would represent a potential more environmentally friendly alternative to chemical synthesis in the oxyfunctionalization reactions described during this Thesis.

1 Introduction

Introduction 1

1. Introduction

1.1. Biocatalytic oxyfunctionalization reactions

The awareness of the depletion of natural resources and the need to preserve them is driving the attention of industries to minimize their contribution to environmental pollution making their processes greener and cleaner whenever possible. In this context, the use of for chemical transformations (biocatalysis) is gaining importance as a more environmental-friendly alternative to traditional synthetic chemistry (Sethi et al., 2008).

The economic and environmental benefits that make biocatalysis attractive include: i) the nature of the biocatalysts, which are biodegradable and essentially nonhazardous and nontoxic, and can be readily produced from renewable resources, and ii) the enzymatic reactions, generally performed selectively under mild conditions (ambient temperature and atmospheric pressure) making biocatalytic processes cost-effective and, therefore, more sustainable (Sheldon and Woodley, 2018).

One of the reasons of the late development of biocatalysis is the reduced accessibility to commercial enzymes, limited mainly to (lipases and proteases) for food and beverage processing and laundry detergents (Sheldon and Woodley, 2018). However, in the last decades, the advances in genetic and protein engineering technologies allow for the easier production of better biocatalysts, usually being able to tailor-design them according to the industrial processes demands (Sethi et al., 2008).

The analysis carried out by Liese et al. (2006) and Straathof (2006) gives an idea of the wide variety of sectors in which biotransformation products are applied, fine chemicals being predominant. Besides the pharma sector, biocatalysis is also gaining importance in food and cosmetic industries in which the ease of transforming natural ingredients plays an important role. One third of the commercialized processes included in this analysis are bioredox reactions, and most of the oxidation reactions were oxyfunctionalizations, probably due to worse chemical alternatives to these processes (Fig. 1).

7 1 Introduction

Alcohol/amine oxidation 22% Hydroxylation 40% Dehydrogenation 5%

Epoxidation 14%

5% 14% Baeyer-Villiger Dihydroxylation oxidation Fig. 1. Distribution of reaction types in biocatalytic oxidation reactions in industrial commercialized processes indicating the oxyfunctionalizations in black letters (adapted from Hollmann et al., 2011).

1.2. Oxyfunctionalization biocatalysts: Fungal peroxygenases

Oxygenases are the most exciting enzymes for synthetic chemistry since they can act into non-activated C-H bonds introducing new functionalities to molecules. In the mechanism of oxyfunctionalization reactions, the oxidation can take place through the reductive activation of molecular oxygen (in case of mono- and di- oxygenases) or already reduced hydrogen peroxide (in case of peroxygenases) and the subsequent insertion into the starting material (Fig. 2) (Dong et al., 2018).

Monooxygenases are the main natural catalysts to mediate oxyfunctionalization reactions, being cytochromes P450 (P450s) the largest and more diverse group (Munro et al., 2007). They use Compound I (oxyferryl heme species) as oxygenating specie, but its formation is complex since they frequently need expensive cofactors (NADPH) and auxiliary enzymes to transform the hydride donation into two successive single electron transfers (Dong et al., 2018).

Fungal peroxygenases, described as unspecific peroxygenases (UPOs; EC.1.11.2.1) in IUBMB Enzyme Nomenclature, share some similarities to the well- known P450s, like the use of Compound I as oxygenating specie but its formation is much simpler since they only require H2O2 for activation. Moreover, UPOs are secreted proteins while P450s are intracellular enzymes, making them far more stable (Fig. 2).

8 Introduction 1

OH OH H H R R’ R R’ R R’ R R’ Peroxygenase Monooxygenase O O R R’ R R’ H O H O R O2 H2O R 2 2 2 R’ + R’ Donorred Donorox Cell (hypha) Cell

Fig. 2. Comparison of the biocatalytic oxyfunctionalization reactions catalyzed by monooxygenases and peroxygenases. Adapted from Dong et al., (2018) and Ullrich and Hofrichter, (2005).

The first peroxygenase of this type was discovered in 2004 in the basidiomycete Agrocybe aegerita and was first described as an haloperoxidase for its ability of oxidizing halides and aryl alcohols (Ullrich et al., 2004) (Fig. 3). One year later, the activity on aromatic compounds led to the change in the name to aromatic peroxygenase (APO, Ullrich et al., 2005) and after the demonstration of hydroxylation activity on aliphatics (Gutiérrez et al., 2011; Peter et al., 2011) the name was finally established as unspecific peroxygenase and was included in the enzyme classification as EC 1.11.2.1. Since then, more peroxygenases were described in other basidiomycetes, such as Coprinellus radians (Anh et al., 2007) and Marasmius rotula (Gröbe et al., 2011) and more recently, in the ascomycete Chaetomium globosum (Kiebist et al., 2017). Their widespread occurrence in the fungal kingdom was demonstrated during the study of basidiomycete and ascomycete genomes, in which several-thousand peroxygenases-type genes were identified (Hofrichter et al., 2015) allowing for the production of recombinant enzymes like those from Coprinopsis cinerea (Babot et al., 2012) and Humicola insolens, heterologously expressed by Novozymes A/S in Aspergillus oryzae.

9 1 Introduction

2011: Activity found on aliphatics 2012: 2004: A. aegerita Compound I 2014: O haloperoxidase (n) HO identified as Heterologous (AaP) oxidizes OH key reactive expression of 1995: halides and aryl intermediate CciUPO in Alkaline alcohols Aspergillus

COOH HOOC lignin O N N - - FeIV Br Br N O N •+ Cys in Agrocybe aegerita 1970 1980 1990 2000 2010

1977: 2005: AaP st - 2012: Crystal 1 mention of + O oxidized also N structure of peroxygenase aromatics activity in pea AaeUPO seeds major form 2009: Main 2011: 3rd AaeAPO UPO in M. (apo1) gen rotula secuenced

Fig. 3. Timeline of the most relevant events related to the history of fungal unspecific peroxygenases.

The sequencing of the main gen of Agrocybe aegerita (apo1) in 2009 and its recombinant expression, together with the publication of the crystal structure of the major form, allowed for genetic and protein engineering tools and the generation of variants by directed evolution or semi-rational design, extending the use of these biocatalysts (Fernández-Fueyo et al., 2016; Molina-Espeja et al., 2014; Molina- Espeja et al., 2015).

The crystal structure of AaeUPO II (Fig. 4A) confirmed the presence of a cysteine residue (Cys36) ligating the heme group (heme thiolate protein) (Piontek et al., 2013). Besides the Cys36, there are also other aminoacids highly conserved in most UPOs (Fig. 4B) like: i) the glutamate (Glu122) and serine (Ser126) coordinating one magnesium ion together with the heme propionate; ii) two prolines (Pro35 and Pro37) exposing the Cys36 to the heme; and iii) a glutamic acid (Glu196) in the heme distal side facilitating the peroxide cleavage (Hofrichter et al., 2015). The residue acting as charge stabilizer of this glutamate (an arginine in AaeUPO and rCciUPO or a histidine in MroUPO) was related to the different behavior of these UPOs towards hydrogen peroxide (Karich et al., 2016).

10 Introduction 1

A

B C ARG-189

ASP-124 GLU-196

GLY-123

CYS-36 GLU-122

PRO-35 PRO-37

Fig. 4. Molecular structure of AaeUPO showing the heme ligated to the Cys36, a molecule of substrate (2/5-hydroxymethylimidazole) and a magnesium ion (in pink color) obtained from PDB 2YOR (A). Aminoacid residues at the of AaeUPO (adapted from Hofrichter et al., 2015) (B). Distal heme-access channel showing the presence of several phenylalanine residues (from Piontek et al., 2013) (C).

Concerning the distal heme-access channel, it is noticeable the predominance of aromatic and few aliphatic residues (Fig. 4C), making the environment rather hydrophobic (Piontek et al., 2013). The heme-access channel architecture is one of the features of UPOs determining the different reactivities and selectivities on different substrates. Recently, the wider access channel of MroUPO compared to AaeUPO (Fig 5A-B) was associated to the Cα-hydroxylating ability on fatty acids (Olmedo et al., 2017) and the ability of CglUPO oxidizing gonane ring of testosterone was thought to be due to an even wider access channel of this enzyme (Kiebist et al., 2017).

11 1 Introduction

A B

Phe191

Ac Phe274 Ac Phe69 Phe199 Phe160

Fig. 5. Comparison of heme-access channel of AaeUPO (A) and MroUPO (B). Adapted from Olmedo et al., (2017).

1.3. Reaction cycle and mechanism of peroxygenases

The proposed catalytic cycle for UPOs is shown in Fig. 6. In the scheme, the peroxygenase and peroxidase pathways are shown. The cycle starts with the activation of the resting state of the enzyme (containing Fe3+ and a porphyrin ring) by peroxide, to yield the oxo-ferryl cation radical complex also known as Compound I. In the peroxygenase pathway, Compound I abstracts one H atom from the substrate (R-H) to yield a radical and ferryl hydroxide complex (Compound II) that rapidly recombines to give the hydroxylated product and the enzyme returns to the initial resting state. If another molecule of peroxide reacts with Compound I, it disproportionates into oxygen and water (` activity´). However, in the peroxidase pathway, Compounds I and II abstract two electrons consecutively from two molecules of substrates (R-OH) forming radicals that can undergo coupling reactions (Hofrichter et al., 2015).

The factors determining the pathway used includes the nature of the UPO and the substrate used, their redox potentials, the reaction pH or the size of the heme channel (Hofrichter et al., 2015). For example, typical substrates of the peroxidase pathway are phenolic compounds, usually described as an undesired reaction in aromatic oxyfunctionalizations by UPOs. On the other hand, rCciUPO showed higher catalase activity compared to AaeUPO and MroUPO and the apparent Km

12 Introduction 1

Peroxygenase H R• O Fe (IV)

S-Cys Compound II

H H R-OH R-H O O H2O2 H2O • + Fe(III) (III) Fe(III) (IV)

S-Cys S-Cys Initial resting state H2O + O2 H2O2 Compound I

• R-OH R-O H + H2O O R-OH R-O• Fe (IV)

S-Cys Compound II Peroxidase

Fe =

Fig 6. Proposed reaction cycle of reactions catalyzed by UPOs showing the peroxygenase and peroxidase pathways (adapted from Horn, 2009).

value for catalase activity increased when veratryl alcohol was added, suggesting competition between the second H2O2 molecule and the substrate (Karich et al., 2016).

13 1 Introduction

To demonstrate the peroxygenative activity of these biocatalysts, additional 18 studies were performed using H2 O2 as co-substrate, to confirm the source of the oxygen atom incorporated into the substrate by the shifts in the mass spectra of the resulting products. These experiments confirmed that the phenolic oxygen incorporated by AaeUPO into aromatic substrates like (R)-2-(4- hydroxyphenoxy)propionic acid (Kinne et al., 2008) and 5-hydroxypropranolol (Kinne et al., 2009b) comes from H2O2. In addition, the incorporation of 69% of 18O into 4-nitrobenzaldehyde demonstrated that the oxygen incorporation takes place in the carbonyl group of products coming from ether cleavage by AaeUPO (Kinne et al., 2008).

Moreover, labeling studies indicated that the oxidation of alcohols to carbonyls and carboxyl groups is produced by successive hydroxylations combined with dehydration steps, as it was observed in the reaction of tetradecanol to form tetradecanoic acid by AaeUPO (Gutiérrez et al, 2011).

1.4. Peroxygenase-catalyzed reactions

Peroxygenases use a variety of peroxides (hydrogen or organic peroxides) as terminal electron acceptors and a large number of electron-donating substrates yielding many products of industrial interest (Martínez et al., 2017). One of the most exciting features of UPOs is the promiscuity of substrates they can accept, selectively catalyzing reactions that are usually challenging in Organic Chemistry.

More than 300 compounds have been proved to be UPOs substrates (Hofrichter and Ullrich, 2014) and their catalytic versatility was clearly pointed out during the oxidation of 23 drugs, including aromatic and aliphatic hydroxylation, O- and N- dealkylations and ester cleavage reactions (Poraj-Kobielska et al., 2011), or in the oxidation of 35 out of 40 compounds (chlorinated benzenes derivatives, halogenated biphenyl ethers, nitroaromatic compounds, polycyclic aromatic hydrocarbons and phthalic acid derivatives) included in the EPA pollutant list (Karich et al., 2017).

Some of the most important oxygenation reactions with representative examples are detailed in the following subsections.

14 Introduction 1

1.4.1. Oxygenation of C-H bonds

Selective activation of C-H bonds is one the most challenging reactions in Organic Chemistry. This reaction was described with UPOs and a wide variety of substrates and, in some cases, with different regioselectivity depending on the UPO used (Table 1).

For example, AaeUPO was shown to catalyze the hydroxylation of linear alkanes (C3-C16) at 2- and 3-positions (Table 1, a) while MroUPO was able to catalyze the hydroxylation at terminal position, an inert position very difficult to oxidize chemically (Olmedo et al., 2016), in addition to position 2. The main products of this reaction were the corresponding acids or ketones (Table 1, b). Interestingly, in the hydroxylation of long chain alkanes by AaeUPO and rCciUPO the regioselectivity depended on the proportion of acetone in the reaction (used for substrate solubilization). With low co-solvent concentration (20%) the main products in the reaction were dihydroxy and hydroxy-keto derivatives, while at higher acetone concentration (40-60%) the monohydroxylated derivatives prevailed (Babot et al., 2012).

Concerning cyclic alkanes (C5-C8), the oxidation of cyclohexane to cyclohexanol or cyclohexanone was also reported for several UPOs (Peter et al., 2015). In general, the main product of the reaction was the monohydroxylated product, while the over-oxidized ketone could be more efficiently produced with MroUPO.

Saturated fatty acids (C12-C18) were also reported to be substrates of peroxygenases, again with different regioselectivity depending on the UPO used. While AaeUPO and rCciUPO formed monohydroxy derivatives at ω-1 and ω-2 positions, the main products of the reaction with MroUPO were dicarboxylic and ω-1 keto-acids (Babot et al., 2012; Olmedo et al., 2016).

UPOs can also catalyze enantioselective hydroxylation reactions like in the alkyl chain of ethylbenzene yielding (R)-1-phenylethanol (>99% ee) (Kluge et al., 2012). This reaction has been widely studied as model reaction for several methods of in situ H2O2 generation (Ni et al., 2015), two-liquid phase system and neat reactions (Fernández-Fueyo et al., 2016). Interestingly, when the ethyl group of ethylbenezene is reduced to a methyl group (i.e. toluene) the hydroxylation takes place in both the aromatic ring and the alkyl chain (Ullrich et al., 2005).

15 1 Introduction

Table 1. Reaction schemes of most representative oxygenation reactions of C-H bonds by fungal peroxygenases

Reaction Examples Refs.

Hydroxylation of aliphatic Peter et al., 2011 Olmedo et al., 2016 C-H bonds

Peter et al., 2014

Babot et al., 2012

Kluge et al., 2012

Babot et al., 2015b

Hydroxylation Kluge et al., 2009 of aromatic C-H bonds Karich et al., 2013

Kinne et al., 2008

Aranda et al., 2010

Hydroxylation Kinne et al., 2009 and cleavage Poraj-Kobielska et al., 2011

Olmedo et al., 2017

16 Introduction 1

Bulkier substrates, such as steroids, are also susceptible of oxygenation by UPOs. AaeUPO, MroUPO and rCciUPO were found to hydroxylate steroids preferentially in the aliphatic chain rather than in the steroidal ring, forming predominantly 25-hydroxyderivatives (Babot et al., 2015). An interesting example of this reaction is the hydroxylation of the secosteroid vitamin D in C25 to form the hydroxy derivative active metabolite (Babot et al., 2015). On the other hand, MroUPO also catalyzes the side-chain removal of corticosteroids when they have both hydroxyacetyl and hydroxyl functionalities at C17 (Ullrich et al., 2018). However, when the hydroxyl is the only functionality present in C17, CglUPO is the only enzyme able to transform this substrate, hydroxylating or epoxidating the steroidal ring (Kiebist et al., 2017).

UPOs also catalyze oxygenation reactions in aromatic rings, whether they are substituted (flavonoids, hydroxy propionic acid, propranolol) or not (benzene, naphtalene) (Table 1). In the case of benzene and naphthalene, the hydroxylation proceeds via hydrolysis of an epoxide intermediate (Kluge et al., 2009). In substrates where both hydroxylations in aromatic or aliphatic carbons are possible, there were found differences again depending on the UPO used. This is the case of fluorene, in which hydroxylation in the aromatic ring is favored by AaeUPO while the hydroxy derivative in the alkylic C9 is the main reaction product with CraUPO (Aranda et al., 2010). Phenolic products produced after aromatic hydroxylation are susceptible to peroxidation reactions (see subsection 3 for more information), forming radicals that can undergo coupling reactions.

The third type of C-H bonds oxygenation includes the hydroxylation and subsequent cleavage reaction, the so-called O- and N-dealkylations (Table 1). This reaction was described for secondary/tertiary amines (Poraj-Kobielska et al., 2011) and aromatic, aliphatic, cyclic and acyclic ethers (Kinne et al., 2009). One especial case of hydroxylation and cleavage reaction is the recently described chain- shortening of carboxylic acids by MroUPO, in which the hydroxylation in Cα position leads to oxo intermediates that decarboxylate in the presence of H2O2 (Olmedo et al., 2017). This reaction occurs in dicarboxylic acids and competes with ω and ω-1 hydroxylation in monocarboxylic acids, in which the selectivity depends on the length of the aliphatic chain.

17 1 Introduction

1.4.2. Oxygenation of C=C bond

UPOs catalyze the oxygenation of double bonds in aromatic and aliphatic hydrocarbons to give reactive epoxide products with different yields depending on the linear or cyclic structure of the substrate (Table 2). When a double bond is present in aliphatic molecules, besides epoxidation, hydroxylation takes place in the allylic position in linear (C3-C8) and cyclic (C6) alkenes (Peter et al., 2013). In the case of limonene, the main products of the reactions with AaeUPO were the epoxide derivatives in the ring and in the side-chain and the hydroxy derivative in the allylic position of the ring, which are interesting products as flavor and fragrance additives.

Table 2. Reaction schemes of most representative oxygenation reactions of C=C bonds by fungal peroxygenases

Reaction Examples Refs.

Epoxidation Peter et al., 2013

Peter et al., 2013

Peter et al., 2013

Kluge et al., 2012

Kluge et al., 2012

Kluge et al., 2012

In aromatic substrates (i.e. styrene derivatives), the epoxidation of the double bond in the aliphatic chain predominates. Interestingly, the configuration of the double bond in methylstyrene determined the selectivity of the reactions, being the cis-isomer completely converted into one epoxide enantiomer (>99% ee) while the

18 Introduction 1

trans-isomer was mainly hydroxylated at the terminal carbon (Kluge et al., 2012). The chiral oxiranes are versatile and important building blocks for a broad range of products (Liese et al., 2006).

As mention above, when the epoxidation takes place in aromatic rings, the unstable epoxide hydrolyzes to form the corresponding phenol (Kluge et al., 2009; Karich et al., 2013).

1.4.3. Oxygenation of alcohols and aldehydes

UPOs can also catalyze the oxygenation of primary and secondary alcohols to aldehydes and ketones, which can subsequently be oxidized to the corresponding carboxylic acids (Table 3). This was one of the first activities discovered for AaeUPO, in which aryl alcohols (anisyl, benzyl and veratryl alcohols) were oxidized to their corresponding aldehydes and carboxylic acids (Ullrich et al., 2004). One of the main spectrophotometric assays to measure UPO activity relies on the oxidation of veratryl alcohol to veratraldehyde, that can be followed by the absorbance at 310 nm (ɛ=9300 M-1s-1).

Table 3. Reaction schemes of most representative oxidation reactions of alcohols and aldehydes by fungal peroxygenases

Reaction Examples Refs.

Alcohols and Ullrich et al., 2004 aldehydes oxidation to ketones and carboxylic Ullrich et al., 2004 acids

Babot et al., 2012; Olmedo et al., 2016

Carro et al., 2015

19 1 Introduction

Alcohols with longer aliphatic chains (1-dodecanol and 1-tetradecanol) are also shown to be oxidized by AaeUPO, rCciUPO and MroUPO producing mainly the carboxylic acid and its hydroxylated derivatives, aldehyde being found only in trace amounts (Babot et al., 2012; Olmedo et al., 2016).

An interesting example of aldehyde oxidation by UPOs is included in the cascade reaction proposed by Carro et al., (2015) combining the aryl-alcohol oxidase (AAO) with AaeUPO for the oxidation of 5-hydroxymethylfurfural (HMF) to produce 2,5-furandicarboxylic acid (FDCA), of interest as platform chemical. In this cascade reaction, AaeUPO uses the H2O2 generated by AAO to transform 2,5- formylfurancarboxylic acid (FFCA) into FDCA.

Simple or polysubstituted phenols can be subjected to peroxidative activity of the enzyme releasing radicals (Karich et al., 2013). Typically, ascorbic acid is added to the reaction to reduce the phenoxyl radicals and to avoid coupling reactions (Scheme 1). Directed evolution was also used to modulate the peroxidative/peroxygenative activity of the UPO in the synthesis of 1-naphtol and 5-hydroxypropanolol (Molina-Espeja et al., 2014; Gómez de Santos et al., 2018).

Scheme 1. Mechanism of ascorbate reduction of phenoxyl radicals formed during phenols peroxidation by unspecific peroxygenases (adapted from Hofrichter et al., 2015).

20 Introduction 1

1.4.4. Oxygenation of heteroatoms

UPOs also catalyze oxygen transfer reactions to heteroatoms like nitrogen and sulfur (Table 4), as it was shown in the hydroxylation of dibenzothiophene (Aranda et al., 2009), thioanisole (Bassanini et al., 2017) and pyridine (Ullrich et al., 2008) by AaeUPO. In contrast to CraUPO that preferably hydroxylated the sulfur atom, AaeUPO hydroxylated the benzene ring of dibenzothiophene.

AaeUPO and CraUPO were also described to catalyze oxidation of halides (Br- and in minor amount Cl-) to hypohalites that can in turn halogenate organic substrates (Ullrich et al., 2005). In contrast, MroUPO and CglUPO did not show halogenating activity (Kiebist et al., 2017).

One electron oxidation on heteroatom different from oxygen was also described in the nitrogen of 2,2′-azino-bis(3-ethylbenzothiazoline-6-sulphonic acid) (ABTS) by all UPOs, a typical reaction to measure peroxidase activity.

Table 4. Reaction schemes of most representative oxygenation reactions on heteroatoms by fungal peroxygenases

Reaction Examples Ref

Oxygen Ullrich et al., 2008 transfer to heteroatom Bassanini et al., 2017

Aranda et al., 2009

Ullrich et al., 2005 Anh et al., 2007

21

2 Objectives

Objectives 2

2. Objectives The main objective of this Thesis was the study of oxyfunctionalization reactions catalyzed by several fungal unspecific peroxygenases (UPOs) such as those from the basidiomycetes Agrocybe aegerita (AaeUPO), Marasmius rotula (MroUPO) and the recombinant enzyme from Coprinopsis cinerea (rCciUPO), and those from the ascomycetes Chaetomium globosum (CglUPO) and the recombinant Humicola insolens (rHinUPO) to produce compounds of industrial interest. The reactions included: hydroxylation of aromatic (trans-stilbene and other stilbenoids) and cyclic aliphatic (α-isophorone) substrates and epoxidation of fatty acids and their methyl esters. The specific objectives were:

 To study the regioselectivity of the different wild and recombinant peroxygenases towards the different substrates, to select the best biocatalyst.

 To confirm the source of oxygen atom incorporation into the substrate by 18 using H2 O2 as UPO co-substrate.

 To perform kinetic analysis of the different substrates oxidation by UPOs, compare their turnovers, affinity and catalytic efficiencies, and select the most efficient biocatalyst.

 To optimize the reaction conditions, including substrate, co-substrate and enzyme concentrations, co-solvent proportions (if necessary) and reaction times.

 To understand the different reactivities observed in UPO reactions through inspection into enzyme structure (heme-access channel and cavity) and computational studies.

25

3 Results and Discussion

General results and discussion 3

3. General Results and Discussion In this Thesis, the oxygenation reactions of trans-stilbene and other stilbenoids, α-isophorone, fatty acids (FAs) and fatty acids methyl esters (FAMEs) catalyzed by the peroxygenases from Agrocybe aegerita (AaeUPO), Marasmius rotula (MroUPO), Chaetomium globosum (CglUPO) and the recombinant enzymes from Coprinopsis cinerea (rCciUPO) and Humicola insolens (rHinUPO) were studied.

3.1. Oxygenation of trans-stilbene and other stilbenoids by UPOs Several stilbenoids, namely trans-stilbene (St), pinosylvin (Pin) and resveratrol (RSV), were tested as substrates of the different UPOs (Scheme 2).

Scheme 2. Reactions of trans-stilbene (St, A-B), pinosylvin (Pin, C-D) and resveratrol (RSV, E) by the different UPOs.

29 3 General results and discussion

Among the different reactions catalyzed by the different UPOs, the hydroxylation of trans-stilbene to produce the 4,4′-dihydroxy-trans-stilbene (DHS) is the most interesting one (Scheme 2A). DHS is a resveratrol analogue described as a potent antioxidant and whose preventive effect of cancer invasion and metastasis has recently been shown (Savio et al., 2016).

3.1.1. Oxygenation of trans-stilbene

Regioselectivity in trans-stilbene oxygenation In the reactions with trans-stilbene, complete substrate conversion (100%) was achieved with all UPOs tested but different regioselectivities were observed (Fig. 7). The peroxygenases from the basidiomycetes (AaeUPO, MroUPO and rCciUPO) selectively hydroxylated the para-positions of the substrate to form DHS. In contrast, the peroxygenase from the ascomycete (CglUPO) did not hydroxylate the aromatic rings and instead formed exclusively the epoxide (trans-stilbene epoxide) in the aliphatic chain of trans-stilbene. Product yields (of both DHS and the epoxide) were over 90% for all the UPOs tested.

St-epoxide DHS St

CglUPO rCciUPO MroUPO AaeUPO Control 15 20 25 Retention time (min)

Fig. 7. GC–MS analysis of trans-stilbene reactions with AaeUPO, rCciUPO, MroUPO and CglUPO and control without enzyme, showing the remaining substrate (St, trans-stilbene), and the dihydroxylated (DHS, 4,4′-dihydroxy-trans-stilbene) and epoxide (St-epoxide; trans-stilbene epoxide) derivatives.

30 General results and discussion 3

In order to confirm the source of the oxygen incorporated during the reaction, 18 18 O-labeling studies were performed with H2 O2 as UPO co-substrate. The mass spectra of DHS and trans-stilbene epoxide from enzymatic reactions showed a shift in the molecular ion fragments, from m/z 356 and 196, respectively, to m/z 360 and 198, confirming the oxygen incorporation from the peroxide and the peroxygenative mechanism of the reaction.

Kinetic studies of trans-stilbene hydroxylation

In order to get additional information between the differences among the different biocatalysts, kinetic parameters for trans-stilbene hydroxylation were estimated (Table 5).

Table 5. Estimated kinetic parameters for trans-stilbene (St) and 4-hydroxy-trans- stilbene (4HS) hydroxylation by unspecific peroxygenases (AaeUPO, MroUPO and rCciUPO). Data represent mean values of three replicates with standard errors.

-1 -1 -1 kcat (s ) Km (µM) kcat/ Km (M ·s ) St→4HS AaeUPO 85.7 ± 8.5 20.8 ± 5.6 4.1 ± 1.2 x 106 MroUPO 4.2 ± 0.2 15.3 ± 2.5 2.7 ± 0.5 x 105 rCciUPO 9.1 ± 0.8 30.5 ± 7.0 3.0 ± 0.7 x 105 4HS→DHS AaeUPO 225 ± 48.3 14.1 ± 7.7 1.6 ± 0.9 x 107 MroUPO 11.7 ± 0.3 46.4 ± 2.3 2.5 ± 0.1 x 105 rCciUPO 6.6 ± 1.3 18.9 ± 8.5 3.5 ± 0.8 x 105

AaeUPO was the most efficient enzyme catalyzing the double hydroxylation of trans-stilbenes. While the Km values were similar for all enzymes, the turnover numbers (kcat) were much higher for AaeUPO, resulting in a 15- and 60-fold increase in the catalytic efficiency (kcat/ Km) for the first and second hydroxylation step, respectively. It was also found that 4-hydroxy-trans-stilbene (4HS) hydroxylation by AaeUPO was more favorable than that of trans-stilbene as it exhibited higher kcat and lower Km values, which increased the catalytic efficiency four times. On the other hand, the catalytic efficiencies were roughly in the same range for both hydroxylation steps in the case of MroUPO and rCciUPO.

31 3 General results and discussion

The catalytic differences between the UPOs oxyfunctionalizating trans-stilbene were supported by their reaction performance. AaeUPO displayed a higher total turnover number (TTN) of 200000, while 26000 and 1400 was attained with MroUPO and rCciUPO, respectively.

3.1.2. Oxygenation of other stilbenoids

Other stilbenoids, such as pinosylvin and resveratrol, were also tested as substrates of the different UPOs. In the reactions with pinosylvin, different conversion rates and product patterns were observed depending on the particular UPO tested (Fig. 8).

A B

RSV oxyRSV

RSV oxyRSV Pin

5 10 15 20 5 10 15 20 Retention time (min) Retention time (min)

C D RSV RSV oxyPin Pin oxyRSV oxyRSV

5 10 15 20 5 10 15 20 Retention time (min) Retention time (min) Fig. 8. Reverse phase HPLC analysis of products from pinosylvin (Pin, A and C) and resveratrol (RSV, B and D) enzymatic reactions with AaeUPO (upper) and rCciUPO (bottom) showing the remaining substrate and the hydroxylated products oxypinosylvin (oxyPin), resveratrol (RSV) and oxyresveratrol (oxyRSV).

32 General results and discussion 3

Using AaeUPO and rCciUPO, higher (97 and 80% respectively) conversions of pinosylvin was attained, while only 11% transformation was produced with MroUPO and no conversion was observed with CglUPO. The enzymatic reactions with pinosylvin and AaeUPO yielded resveratrol that was in turn oxidized to oxyresveratrol; by contrast rCciUPO produced, in addition, another monohydroxylated compound that was tentatively identified as 2′- hydroxypinosylvin. This suggests that AaeUPO seems to be more selective towards the position 4′ of pinosylvin, while rCciUPO hydroxylates both positions (2′ and 4′) to similar extent. When resveratrol was used as substrate, up to 72% and 22% conversion into oxyresveratrol was attained with AaeUPO and rCciUPO, respectively, but no conversion was observed with CglUPO and MroUPO.

3.1.3. Structure-activity analysis

The different reactivity of UPOs towards trans-stilbene, pinosylvin and resveratrol could be attributable to the differences found in their structures. Analyzing the heme-access channel of UPOs, the main difference was found with respect to the number of solvent-exposed phenyalanine residues. The presence of four phenyalanines at the entrance of the heme-access channel in AaeUPO making it more affine to aromatics (compared to MroUPO and rCciUPO that have only one) might be the reason of substrate positioning with the para-position close to the heme.

Interestingly, if the heme-cavity is inspected, CglUPO has the most different active site compared to the rest of UPOs (Fig. 9). It has only in common with AaeUPO the glutamic acid residue (putatively involved in H2O2 cleavage) that is presumably stabilized by a histidine, like MroUPO and in contrast to AaeUPO and rCciUPO bearing an arginine in this position. Another interesting peculiarity is the presence of a reactive tyrosine (Tyr162) in the middle of the active site, which is only present in this UPO.

33 3 General results and discussion

A B

T192 F199 I153 Ac H86 R189 Ac F160 E196 E157

I84 F121 C36 C17

C D F154 T192 F199

R189 H88 Y162 E196 E158

L86 F121 C36 C17

Fig. 9. Residues present in the active site of AaeUPO (A), MroUPO (B), rCciUPO (C) and CglUPO (D) showing in black the same present in AaeUPO while varying residues have different colors.

3.2. Oxygenation of α-isophorone by UPOs The biocatalytic oxygenation of α-isophorone to form the compounds of industrial interest 4-hydroxyisophorone (4HIP) and 4-ketoisophorone (4KIP) was studied using several UPOs (Scheme 4). 4HIP and 4KIP are used as flavour and fragrance additives and as intermediates in the synthesis of pharmaceuticals, vitamins and natural pigments (Krill et al., 2002; Eggersdorfer et al., 2012; Isler et al., 1956).

34 General results and discussion 3

Scheme 4. Enzymatic hydroxylation of α -isophorone (IP) by CglUPO, HinUPO (A) and AaeUPO (B) forming 4-hydroxyisophorone (4HIP), 4-ketoisophorone (4KIP), 7- hydroxyisophorone (7HIP) and 7-formylisophorone (7FIP).

3.2.1. Regioselectivity in α-isophorone hydroxylation

The peroxygenases from the ascomycetes (CglUPO and rHinUPO) were found to selectively hydroxylate α-isophorone to form the compounds of interest 4HIP and 4KIP, while the peroxygenase from the basidiomycete (AaeUPO) formed in addition to 4HIP, the oxidized derivatives 7-hydroxyisophorone (7HIP) and 7- formylisophorone (7FIP). Only traces of 4KIP were found in the last reaction. In contrast, MroUPO and rCciUPO were unable to transform the substrate (Fig. 10).

35 3 General results and discussion

IP

4HIP 7FIP 7HIP 4KIP AaeUPO rHinUPO CglUPO rCciUPO MroUPO Control 6 7 8 9 10 11 12 Retention time (min) Fig. 10. GC-MS analysis of products from α-isophorone (IP) enzymatic reaction (30 min) with several peroxygenases showing the different products: 4-hydroxyisophorone (4HIP), 4-ketoisophorone (4KIP), 7-hydroxyisophorone (7HIP) and 7-formylisophorone (7FIP).

Reactions with higher substrate load (10 mM) were performed with CglUPO and rHinUPO, selected for their higher selectivity (Fig. 11). rHinUPO was found to completely convert the substrate in 6 h, while 87% conversion was achieved with CglUPO within 12 hours.

10 B 10 IP A 4-HIP 4-KIP

8 8

)

)

mM

mM

( ( 6 6

4 4

Concentration Concentration 2 2

0 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Time (h) Time (h) Fig. 11. Time course of the isophorone (10 mM) enzymatic reactions with CglUPO (A) and rHinUPO (B) showing the remaining substrate and the products 4-hydroxyisophorone (4HIP) and 4-ketoisophorone (4KIP).

36 General results and discussion 3

Higher amounts of the keto-derivative were found in the reactions with rHinUPO, while 4HIP was the main product formed with CglUPO. By contrast, functionally similar cytochromes P450 seemed unable to oxidize 4HIP into 4KIP and two enzymes were needed to catalyze the double oxidation of IP to KIP (Tavanti et al., 2017). The study of the enantioselectivity of the synthesis of 4HIP revealed that the reaction with AaeUPO was highly stereo-selective, with an enantiomeric excess (ee) of 88%, while this value was only 40% and 4% with CglUPO and rHinUPO, respectively. Interestingly, all UPOs formed preferentially the S-enantiomer, in contrast to cytochromes P450, that preferentially formed the R-enantiomer (Tavanti et al., 2017; Kaluzna et al., 2016). When the racemic mixture of 4HIP was used as substrate, a kinetic resolution of the racemate was produced with higher velocity in the conversion of the S- enantiomer with CglUPO and rHinUPO (Fig. 12). High ee (99-100%) and 60-75% recovery of the R-enantiomer was achieved within 15-60 min of reaction. However, AaeUPO converted the racemic mixture of 4HIP very slowly and no stereoselectivity was observed.

A B (S) (R) 100 100

(%) 80 80 (%)

60 60

substrate substrate

40 40

20 20

Remaining Remaining

0 0 0 5 15 30 60 0 5 15 30 60 Time (min) Time (min) Fig. 12. Different proportions of R- and S- enantiomers during the course of the enzymatic reaction with CglUPO (A) and rHinUPO (B) when racemic 4HIP (10 mM) was used as substrate.

37 3 General results and discussion

3.2.2. Kinetic studies of α-isophorone hydroxylation

A kinetic study of the α-isophorone hydroxylation by the different UPOs was performed and the catalytic constants (kcat and Km) were estimated. The main difference found in the kinetic studies of α-isophorone hydroxylation by UPOs was the enzyme affinity (Fig. 13). CglUPO and rHinUPO displayed a four- and two- fold lower Km values (309 and 633 µM, respectively) than AaeUPO (1380 µM), revealing higher substrate affinity. Moreover, rHinUPO displayed a ten-fold higher -1 turnover number (kcat, 42 s ) which resulted in five-fold higher catalytic efficiency -1 -1 (kcat/ Km, 66.4 mM s ), while the catalytic efficiencies of the other two enzymes were similar (14.2-14.6 mM-1s-1).

4 20 A B

3 15

) )

-1 -1

2 10

Turnover (s Turnover Turnover (s Turnover 1 5

0 0 0 1000 2000 3000 0 200 400 600 800 1000 1200 1400 1600 1800 IP (µM) IP (µM)

20 18 C 16

) 14

-1 12 10 8

Turnover (s Turnover 6 4 2 0 0 2000 4000 6000 IP (µM) Fig. 13. Kinetic curves of enzymatic hydroxylation of α-isophorone (IP) by CglUPO (A), rHinUPO (B) and AaeUPO (C) from GC-MS estimation of 4HIP/4KIP formation (initial rates).

38 General results and discussion 3

3.2.3. Computational analysis

The experimental results that show different reactivities in α-isophorone hydroxylation by the different UPOs could be rationalized by computational analysis. On one hand, the three UPOs (CglUPO, AaeUPO and rHinUPO) that were found to transform the substrate (Fig. 10) present catalytic distances (<3 Å) between C4 of α-isophorone and the heme-oxo (Compound I) (Fig. 14A).

A

) mol

CglUPO

M roUPO Binding energy (kcal/Bindingenergy AaeUPO rHinUPO rCciUPO

IP-C4 to heme-oxo distance (Å)

B

) mol

CglUPO Binding energy (kcal/Bindingenergy AaeUPO

HIP-C4 to heme-oxo distance (Å) Fig. 14. Results from α-isophorone (IP) and 4-hydroxyisophorone (4HIP) diffusion refinement on five UPOs with adaptive PELE monitoring the distance between the molecule C4 and H4 with the oxo atom of the Compound I with respect to the binding energy. C4 distance vs energy plot for IP diffusion in CglUPO (red), MroUPO (magenta), AaeUPO (green), rHinUPO (cyan) and rCciUPO (blue) (A). H4 distance vs energy plot for S-4HIP diffusion on CglUPO (red) and AaeUPO (green) (B).

39 3 General results and discussion

In addition, a closer examination of α-isophorone position in CglUPO and AaeUPO, showed shorter distances to the heme-oxo for the pro-S (1.9 Å) hydrogen than the pro-R (3.5 and 2.6 Å, respectively), explaining the preferential S- enantiomer of 4HIP formation. On the other hand, strong differences were also found between CglUPO and AaeUPO in the distances with the H4 of the resulting 4HIP and the heme-oxo (Fig. 14B). The fact that no catalytic distances were obtained with AaeUPO is in agreement with the experimental lack of oxidizing activity on 4HIP, as only traces of 4KIP were found in the reactions (Fig. 10).

3.3. Epoxidation of unsaturated fatty acids and methyl esters by UPOs

The epoxidation of unsaturated fatty acids (FAs) and fatty acid methyl esters (FAMEs) was studied with CglUPO and MroUPO to produce reactive epoxides to be used in industrial synthesis of chemicals and intermediates (Scheme 5). Unsaturated FAs are available from vegetable oils by simple industrial operation, and their methyl esters are usually obtained by transesterification of these oils with methanol.

Scheme 5. General scheme of the epoxidation of unsaturated fatty acids (R=H) and their methyl esters (R=CH3) by CglUPO and MroUPO.

3.3.1. Regioselectivity of the epoxidation of unsaturated fatty acids and fatty acid methyl esters

The oxidation of cis-monounsaturated (from C14:1 to C22:1) and polyunsaturated linoleic (cis,cis-9,12-octadecadienoic acid), α-linolenic (cis,cis,cis-9,12,15-octadecatrienoic acid) and γ-linolenic (cis,cis,cis-6,9,12-

40 General results and discussion 3

octadecadienoic acid) acid was accomplished with CglUPO and MroUPO, attaining conversions >75% in all cases (Fig. 15A-D). In general terms, CglUPO was found to be highly selective to form mono- and di-epoxides, while the selectivity of MroUPO depended on the FA chain length. Interestingly, only one monoepoxide product was observed in the reactions with γ-linolenic acid, while two monoepoxy- and di-epoxy derivatives were formed in α-linolenic acid reactions, under the reaction conditions used (Fig. 15C-D).

A B di-E-syn E di-E-anti

C18:1 cis-Δ9 ED C18:2 cis,cis-Δ9,Δ12 18 20 22 24 26 28 30 18 20 22 24 26 28 30 Retention time (min) Retention time (min)

C D E di-E E

(w-4)-OH

α-C18:3 γ-C18:3

18 20 22 24 26 18 20 22 24 26 Retention time (min) Retention time (min)

Fig. 15. GC-MS analysis of the products formed in the enzymatic reactions of CglUPO with the unsaturated oleic acid (A), linoleic acid (B), α-linolenic acid (C) and γ-linolenic acid (D), showing the formation of epoxides (E), diepoxides (di-E), epoxidated derivatives (ED) and the hydroxylated derivative in ω-4 position (ω-4-OH).

41 3 General results and discussion

The highest epoxidation selectivity of CglUPO was observed in the reaction with oleic acid (Fig. 16) in which a double dose of MroUPO was needed to achieve similar conversion rates. Moreover, MroUPO formed additional hydroxy- and keto- derivatives of FA and their epoxides.

A B

Fig. 16. Relative percentage of products from oleic acid enzymatic reaction with CglUPO (A, 50 nM) and MroUPO (B, 100 nM) using 5 mM of H2O2 as co-substrate.

CglUPO and MroUPO also oxidized methyl esters of oleic and linoleic acid (Fig.17A-B) but, in contrast to MroUPO, higher doses of CglUPO were necessary to achieve conversions comparable to that obtained with the respective free FA. CglUPO showed similar selectivity to form the epoxides than with free FA, while the selectivity of MroUPO to form the epoxide of methyl oleate increased compared to the free oleic acid.

This regioselectivity found with CglUPO and MroUPO is different from that found with the previously studied peroxygenases AaeUPO and rCciUPO, where hydroxylation at subterminal positions (ω-1 and ω-2) occurred in both, saturated and unsaturated FAs and FAMEs (Babot et al., 2012).

42 General results and discussion 3

A B

di-E-anti E

di-E-syn

9 CH3-C18:1cis-Δ 9 12 CH3-C18:2 cis,cis-Δ ,Δ

18 20 22 24 18 20 22 24 Retention time (min) Retention time (min) Fig. 17. GC-MS analysis of the products formed in the enzymatic reactions of CglUPO with methyl oleate (A) and methyl linoleate (B), showing the formation of epoxide (E) and diepoxides (di-E).

3.3.2. Kinetic studies of fatty acid oxygenation The higher efficiency of CglUPO oxidizing unsaturated FAs was supported by kinetic studies (Table 6), although difficulties in the calculations were found due to substrate solubilization limitations. CglUPO presented one order of magnitude higher catalytic efficiency (kcat/Km) due to the three-fold higher turnover number (kcat) and four-fold lower Km, representing higher affinity than MroUPO.

Table 6. Estimated kinetic parameters for unsaturated fatty acids oxidation by MroUPO and CglUPO. Data represent mean values of three replicates with standard errors -1 -1 -1 kcat (s ) Km (µM) kcat/ Km (M ·s ) MroUPO Oleic acid 2.6 ± 0.2 38.9 ± 6.1 6.7 ± 1.1 x 104 Linoleic acid 7.8 ± 1.5 372.5 ± 117.9 2.1 ± 0.8 x 104 CglUPO Oleic acid 8.1 ± 0.9 10.7 ± 4.0 7.6 ± 3.0 x 105 Linoleic acid 29.9 ± 2.4 96.7 ± 20.0 3.1 ± 0.7 x 105

On the other hand, CglUPO also displayed up to 8000 total turnover numbers (TTN) and 2.2 s-1 turnover frequency (TOF) when the substrate concentration was increased to 1 mM, while these values were half for MroUPO.

43 3 General results and discussion

In summary, the UPOs studied are able to catalyze the selective oxygenation of trans-stilbene and other stilbenoids, α-isophorone and unsaturated FAs and FAMEs under mild and environmentally friendly reaction conditions, presenting each UPO different regioselectivity and catalytic efficiency. These biocatalysts are extracellular enzymes and use H2O2 as the only co-substrate, catalyzing the insertion of an oxygen atom into the substrate and releasing H2O as by-product. For these reasons, UPOs are appealing biocatalysts for synthetic chemistry and should be considered for enzymatic oxyfunctionalization reactions.

44

4 References

References 4

4. References Anh DH, Ullrich R, Benndorf D, Svatos A, Muck A, Hofrichter M. 2007. The coprophilous mushroom Coprinus radians secretes a haloperoxidase that catalyzes aromatic peroxygenation. Applied and Environmental Microbiology 73: 5477–5485. Aranda E, Kinne M, Kluge M, Ullrich R, Hofrichter M. 2009. Conversion of dibenzothiophene by the mushrooms Agrocybe aegerita and Coprinellus radians and their extracellular peroxygenases. Applied Microbiology and Biotechnology 82: 1057–1066. Aranda E, Ullrich R, Hofrichter M. 2010. Conversion of polycyclic aromatic hydrocarbons, methyl naphthalenes and dibenzofuran by two fungal peroxygenases. Biodegradation 21: 267–281. Babot ED, del Río JC, Kalum L, Martínez AT, Gutiérrez A. 2012. Oxyfunctionalization of aliphatic compounds by a recombinant peroxygenase from Coprinopsis cinerea. Biotechnology and Bioengineering 110: 2323– 2332. Babot ED, del Río JC, Cañellas M, Sancho F, Lucas F, Guallar V, Kalum L, Lund H, Gröbe G, Scheibner K, Ullrich R, Hofrichter M, Martínez AT, Gutiérrez A. 2015. Steroid hydroxylation by basidiomycete peroxygenases: A combined experimental and computational study. Applied and Environmental Microbiology 81: 4130–4142. Babot ED, del Río JC, Kalum L, Martínez AT, Gutiérrez A. 2015b. Regioselective hydroxylation in the production of 25-hydroxyvitamin D by Coprinopsis cinerea peroxygenase. ChemCatChem 7: 283–290. Bassanini I, Ferrandi EE, Vanoni M, Ottolina G, Riva S, Crotti M, Brenna E, Monti D. 2017. Peroxygenase-catalyzed enantioselective sulfoxidations. European Journal of Organic Chemistry 2017; 7186–7189. Carro J, Ferreira P, Rodríguez L, Prieto A, Serrano A, Balcells B, Ardá A, Jiménez- Barbero J, Gutiérrez A, Ullrich R, Hofrichter M, Martínez A. 2015. 5- hydroxymethylfurfural conversion by fungal aryl-alcohol oxidase and unspecific peroxygenase. FEBS Journal 282: 3218–3229. Dong J, Fernández-Fueyo E, Hollmann F, Paul CE, Pesic M, Schmidt S, Wang Y, Younes S, Zhang W. 2018. Biocatalytic oxidation reactions: a chemist’s perspective. Angewandte Chemie International Edition 57: 9238–9261.

49 4 References

Fernández-Fueyo E, Ni Y, Gomez-Baraibar A, Alcalde M, M. van Langenc L, Hollmann F. 2016. Towards preparative peroxygenase-catalyzed oxyfunctionalization reactions in organic media. Journal of Molecular Catalysis B: Enzymatic 134: 347–352. Gómez de Santos P, Cañellas M, Tieves F, Younes SHH, Molina-Espeja P, Hofrichter M, Hollmann F, Guallar V, Alcalde M. 2018. Selective synthesis of the human drug metabolite 5′-hydroxypropranolol by an evolved self- sufficient peroxygenase. ACS Catalysis 8 (6): 4789–4799. Gröbe G, Ullrich M, Pecyna M, Kapturska D, Friedrich S, Hofrichter M, Scheibner K. 2011. High-yield production of aromatic peroxygenase by the agaric fungus Marasmius rotula. Applied and Industrial Microbiology and Biotechnology Express 1: 31–42. Gutiérrez A, Babot ED, Ullrich R, Hofrichter M, Martínez AT, del Río JC. 2011. Regioselective oxygenation of fatty acids, fatty alcohols and other aliphatic compounds by a basidiomycete heme-thiolate peroxidase. Archives of Biochemistry and Biophysics 514: 33–43. Eggersdorfer M, Laudert D, Létinois U, McClymont T, Medlock J, Netscher T, Bonrath W. 2012. One hundred years of vitamins-A success story of the natural sciences. Angewandte Chemie International Edition 51: 12960–12990. Isler O, Lindlar H, Montavon M, Ruegg R, Saucy G, Zeller P. 1956. Synthesen in der carotinoid‐reihe mitteilung. Totalsynthese von zeaxanthin und physalien. Helvetica Chimica Acta 39: 2041–2053. Hofrichter M, Ullrich R. 2014. Oxidations catalyzed by fungal peroxygenases. Current Opinion in Chemical Biology 19: 116–125. Hofrichter M, Kellner H, Pecyna MJ, Ullrich R. 2015. Fungal unspecific peroxygenases: hemo-thiolate proteins that combine peroxidase and cytochrome P450 properties. Advances in Experimental Medicine and Biology 851: 341–368. Hollmann F, Arends I, Buehler K, Schallmey A, Büuhler B. 2011. Enzyme- mediated oxidations for the chemist. Green Chemistry 13: 226–265. Horn A. 2009. The use of a novel peroxidase from the basidiomycete Agrocybe aegerita as an example of enantioselective sulfoxidation. Ph. D. Thesis, University of Rostock, Germany. Kaluzna I, Schmitges T, Straatman H, van Tegelen D, Müller D, Schürmann M, Mink D. 2016. Enabling selective and sustainable P450 oxygenation

50 References 4

technology. Production of 4-hydroxy-α-isophorone on kilogram Scale. Organic Process Research and Development 20 (4): 814–819. Karich A, Kluge M, Ullrich R, Hofrichter M. 2013. Benzene oxygenation and oxidation by the peroxygenase of Agrocybe aegerita. Applied and Industrial Microbiology and Biotechnology Express 3: 5. Karich A, Scheibner K, Ullrich R, Hofrichter M. 2016. Exploring the catalase activity of unspecific peroxygenases and the mechanism of peroxide- dependent heme destruction. Journal of Molecular Catalysis B: Enzymatic 134: 238–246. Karich A, Ullrich R, Scheibner K, Hofrichter M. 2017. Fungal unspecific peroxygenases oxidize the majority of organic EPA priority pollutants. Frontiers in Microbiology 8: 1463. Kiebist J, Schmidtke K, Zimmermann J, Kellner H, Jehmlich N, Ullrich R, Zänder D, Hofrichter M, Scheibner K. 2017. A peroxygenase from Chaetomium globosum catalyzes the selective oxygenation of testosterone. ChemBioChem 18: 563–569. Kinne M, Ullrich R, Hammel KE, Scheibner K, Hofrichter M. 2008. Regioselective preparation of (R)-2-(4-hydroxyphenoxy) propionic acid with a fungal peroxygenase. Tetrahedron Letters 49: 5950–5953. Kinne M, Poraj-Kobielska M, Ralph SA, Ullrich R, Hofrichter M, Hammel KE. 2009. Oxidative cleavage of diverse ethers by an extracellular fungal peroxygenase. The Journal of Biological Chemistry 284: 29343–29349. Kinne M, Poraj-Kobielska M, Aranda E, Ullrich R, Hammel KE, Scheibner K, Hofrichter M. 2009b. Regioselective preparation of 5-hydroxypropranolol and 4-hydroxydiclofenac with a fungal peroxygenase. Bioorganic & Medicinal Chemistry Letters 19: 3085–3087. Kluge M, Ullrich R, Dolge C, Scheibner K, Hofrichter M. 2009. Hydroxylation of naphthalene by aromatic peroxygenase from Agrocybe aegerita proceeds via oxygen transfer from H2O2 and intermediary epoxidation. Applied Microbiology and Biotechnology 81: 1071–1076. Kluge M, Ullrich R, Scheibner K, Hofrichter M. 2012. Stereoselective benzylic hydroxylation of alkylbenzenes and epoxidation of styrene derivatives catalyzed by the peroxygenase of Agrocybe aegerita. Green Chemistry 14: 440–446.

51 4 References

Krill S, Huthmacher K and Perrin S. 2002. Process for the production of an intermediate mixture of β-isophorone epoxide and its isomer 4- hydroxyisophorone. US Patent 6469215 B1. Liese A, Seelbach K, and Wandrey C. 2006. Industrial Biotransformations. 2nd Edition. Wiley-VCH, Weinheim. 425 pp. ISBN: 3-527-30094-5. Martínez AT, Ruiz-Dueñas FJ, Camarero S, Serrano A, Linde D, Lund H, Vind J, Tovborg M, Herold-Majumdar OM, Hofrichter M, Liers C, Ullrich R, Scheibner K, Sannia G, Piscitelli A, Pezzella C, Sener ME, Kılıç S, van Berkel WJH, Guallar V, Lucas MF, Zuhse R, Ludwig R, Hollmann F, Fernández- Fueyo E, Record E, Faulds EB, Tortajada M, Winckelmann I, Rasmussen J, Gelo-Pujic M, Gutiérrez A, del Río JC, Rencoret J, Alcalde M. 2017. Oxidoreductases on their way to industrial biotransformations. Biotechnology Advances 35 (6): 815–831. Molina-Espeja P, García-Ruiz E, González-Pérez D, Ullrich R, Hofrichter M, Alcalde M. 2014. Directed evolution of unspecific peroxygenase from Agrocybe aegerita. Applied and Environmental Microbiology 80 (11): 3496– 350. Molina-Espeja P, Mab S, Matec DM, Ludwig R, Alcalde M. 2015. Tandem-yeast expression system for engineering and producing unspecific peroxygenase. Enzyme and Microbial Technology 73–74: 29–33. Munro, A W, Girvan, H M, and McLean, K J. 2007. Variations on a (t) heme – novel mechanisms, redox partners and catalytic functions in the cytochrome P450 superfamily. Natural Product Reports 24: 585–609. Ni Y, Fernández-Fueyo E, Gomez Baraibar A, Ullrich R, Hofrichter M, Yanase H, Alcalde M, van Berkel WJH, Hollmann F. 2012. Peroxygenase-catalyzed oxyfunctionalization reactions promoted by the complete oxidation of methanol. Angewandte Chemie International Edition 51(52):12960–12990. Olmedo A, Aranda C, del Río JC, Kiebist J, Scheibner K, Martínez AT, Gutiérrez A. 2016. From alkanes to carboxylic acids: Terminal oxygenation by a fungal peroxygenase. Angewandte Chemie International Edition 55: 12248–12251. Olmedo A, del Río JC, Kiebist J, Ullrich R, Hofrichter M, Scheibner K, Martínez AT, Gutiérrez A. 2017. Fatty acid chain shortening by a fungal peroxygenase. Chemistry – A European Journal 23: 16985–16989.

52 References 4

Peter S, Kinne M, Wang X, Ullrich R, Kayser G, Groves JT, Hofrichter M. 2011. Selective hydroxylation of alkanes by an extracellular fungal peroxygenase. The FEBS Journal 278: 3667–3675. Peter S, Kinne M, Ullrich R, Kayser G, Hofrichter M. 2013. Epoxidation of linear, branched and cyclic alkenes catalyzed by unspecific peroxygenase. Enzyme and Microbial Technology 52: 370–376. Peter S, Karich S, Ullrich R, Gröbe G, Scheibner K, Hofrichter M. 2014. Enzymatic one-pot conversion of cyclohexane into cyclohexanone: Comparison of four fungal peroxygenases. Journal of Molecular Catalysis B: Enzymatic 103: 47– 51. Piontek K, Strittmatter E, Ullrich R, Gröbe G, Pecyna MJ, Kluge M, Scheibner K, Hofrichter M, Plattner DA. 2013. Structural basis of substrate conversion in a new aromatic peroxygenase: P450. Functionality with benefits. The Journal of Biological Chemistry 288: 34767–34776. Poraj-Kobielska M, Kinne M, Ullrich R, Scheibner K, Kayser G, Hammel KE, Hofrichter M. 2011. Preparation of human drug metabolites using fungal peroxygenases. Biochemical Pharmacology 82: 789–796. Savio M, Ferraro D, Maccario C, Vaccarone R, Jensen LD, Corana F, Mannucci B, Bianchi L, Cao Y, Stivala LA. 2016. Resveratrol analogue 4,4′-dihydroxy- trans-stilbene potently inhibits cancer invasion and metastasis. Scientific Reports 6:19973. Sethi MK, Chakraborty P, Shukla R. 2008. Biocatalysis: An Industrial Perspective. Royal Society of Chemistry. ISBN: 978-1-78262-619-0. Sheldon RA, Woodley JM. 2018. Role of biocatalysis in sustainable chemistry. Chemical Reviews 118(2):801–838. Straathof A. 2006. Quantitative analysis of industrial biotransformation. In: Industrial Biotransformations (Eds. A. Liese, K. Seelbach, C. Wandrey). 2nd Edition. Wiley-VCH, Weinheim. 515-520. Tavanti M, Parmeggiani F, Gómez Castellanos JR, Mattevi A, Turner NJ. 2017. One‐pot biocatalytic double oxidation of α‐isophorone for the synthesis of ketoisophorone. ChemCatChem. 9: 3338 –3348. Ullrich R, Nuske J, Scheibner K, Spantzel J, Hofrichter M. 2004. Novel haloperoxidase from the agaric basidiomycete Agrocybe aegerita oxidizes aryl

53 4 References

alcohols and aldehydes. Applied and Environmental Microbiology 70: 4575– 4581. Ullrich R, Hofrichter M. 2005. The haloperoxidase of the agaric fungus Agrocybe aegerita hydroxylates toluene and naphthalene. FEBS Letters 579: 6247–6250. Ullrich R, Dolge C, Kluge M, Hofrichter, M. 2008. Pyridine as novel substrate for regioselective oxygenation with aromatic peroxygenase from Agrocybe aegerita. FEBS Letters 582: 4100–4106. Ullrich R, Poraj-Kobielska M, Scholze S, Halbout C, Sandvoss M, Pecyna MJ, Scheibner K, Hofrichter M. 2018. Side chain removal from corticosteroids by unspecific peroxygenase. Journal of Inorganic Biochemistry 183: 84–93.

54

5 Publications

Publications and Patents 5

5. Publications

The main results obtained during this Thesis were published in the following scientific articles and were the object of two patents:

5.1. Scientific articles

 Publication I: Aranda C., Ullrich R., Kiebist J., Scheibner K., del Río J.C., Hofrichter M., Martínez A.T. and Gutiérrez A. (2018). Selective synthesis of the resveratrol analogue 4,4′-dihydroxy-trans-stilbene and stilbenoids modification by fungal peroxygenases. Catalysis Science & Technology 8: 2394–2401.  Publication II: Aranda C., Municoy M., Guallar V., Kiebist J., Scheibner K., Ullrich R., del Río J.C., Hofrichter M., Martínez A.T. and Gutiérrez A. (2019). Selective synthesis of 4-hydroxyisophorone and 4-ketoisophorone by fungal peroxygenases. Catalysis Science and Technology (Published online). DOI: 10.1039/C8CY02114G.  Publication III: Aranda C., Olmedo A., Kiebist J., Scheibner K., del Río J.C., Martínez A.T. and Gutiérrez A. (2018). Selective epoxidation of fatty acids and fatty acid methyl esters by fungal peroxygenases. ChemCatChem 10: 3964– 3968.

5.2. Patents

 Patent I: Aranda Oliden C., del Río Andrade J. C., Martínez Ferrer A. T., Gutiérrez Suárez A. (2018). Process for the selective synthesis of 4- hydroxyisophorone and 4-ketoisophorone by fungal peroxygenases. EP 18382872.  Patent II: Fernández Fueyo E., Aranda Oliden C., Gutiérrez Suárez A., Martínez Ferrer A. T. (2018). Method of heterologous expression of active fungal unspecific peroxygenase in bacterial host cells for fatty-acid epoxidation and other oxygenation reactions. EP 18382514.

59

Publications and Patents 5

H O 2 2 H2O

5.1. Publication I: Aranda C., Ullrich R., Kiebist J., Scheibner K., del Río J.C., Hofrichter M., Martínez A.T. and Gutiérrez A. (2018). Selective synthesis of the resveratrol analogue 4,4′-dihydroxy-trans-stilbene and stilbenoids modification by fungal peroxygenases. Catalysis Science & Technology 8: 2394–2401.

61

Publications and Patents 5

Selective synthesis of the resveratrol analogue 4,4′-dihydroxy- trans-stilbene and stilbenoids modification by fungal peroxygenases

Carmen Aranda,a René Ullrich,b Jan Kiebist,c Katrin Scheibner,c José C. del Río,a Martin Hofrichter,b Angel T. Martínez d and Ana Gutiérrez *a

a Instituto de Recursos Naturales y Agrobiología de Sevilla, CSIC, Reina Mercedes 10, E-41012 Seville, Spain. E-mail: [email protected] b Department of Bio- and Environmental Sciences, TU Dresden, Markt 23, 02763 Zittau, Germany c JenaBios GmbH, Löbstedter Str. 80, 07749 Jena, Germany d Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu 9, E-28040 Madrid, Spain

* Corresponding author

Abstract:

This work gives first evidence that the unspecific peroxygenases (UPOs) from the basidiomycetes Agrocybe aegerita (AaeUPO), Coprinopsis cinerea (rCciUPO) and Marasmius rotula (MroUPO) are able to catalyze the regioselective hydroxylation of trans-stilbene to 4,4′-dihydroxy-trans-stilbene (DHS), a resveratrol (RSV) analogue whose preventive effects on cancer invasion and metastasis have very recently been shown. Nearly complete transformation of substrate (yielding DHS) was achieved with the three enzymes tested, using H2O2 as the only co-substrate, with AaeUPO showing exceptionally higher total turnover number (200 000) than MroUPO (26 000) and rCciUPO (1 400). Kinetic studies demonstrated that AaeUPO was the most efficient enzyme catalyzing stilbene dihydroxylation with catalytic efficiencies (kcat/Km) one and two orders of magnitude higher than those of MroUPO and rCciUPO, so that 4-hydroxystilbene appears to be the best UPO substrate reported to date. In contrast, the peroxygenase

63 5 Publications and Patents

from the ascomycete Chaetomium globosum (CglUPO) failed to hydroxylate trans- stilbene at the aromatic ring and instead produced the trans-epoxide in the alkenyl moiety. In addition, stilbenoids such as pinosylvin (Pin) and RSV were tested as substrates for the enzymatic synthesis of RSV from Pin and oxyresveratrol (oxyRSV) from both RSV and Pin. Overall, lower conversion rates and regioselectivities compared with trans-stilbene were accomplished by three of the UPOs, and no conversion was observed with CglUPO. The highest amount of RSV (63% of products) and oxyRSV (78%) were again attained with AaeUPO. True 18 18 peroxygenase activity was demonstrated by incorporation of O from H2 O2 into the stilbene hydroxylation products. Differences in the number of phenylalanine residues at the heme access channels seems related to differences in aromatic hydroxylation activity, since they would facilitate substrate positioning by aromatic-aromatic interactions. The only ascomycete UPO tested (that of C. globosum) turned out to have the most differing active site (distal side of heme cavity) and reactivity with stilbenes resulting in ethenyl epoxidation instead of aromatic hydroxylation. The above oxyfunctionalizations by fungal UPOs represent a novel and simple alternative to chemical synthesis for the production of DHS, RSV and oxyRSV.

Introduction

Resveratrol (RSV, 3,5-4′-trihydroxy-trans-stilbene) has emerged in recent years as a fascinating compound because of its wide spectrum of biological effects including protection against metabolic, cardiovascular and other age-related complications such as neurodegeneration and cancer.1 Therefore, a large number of structure–activity studies have been carried out that have revealed the molecular determinants of RSV necessary for its biological effects, such as the hydroxyl group in position 4′ of the aromatic ring, together with the trans conformation.2 This information was used for the synthesis of RSV analogues such as 4,4′-dihydroxy- trans-stilbene (DHS) with enhanced cytotoxic, anti-proliferative and anti-tumor properties in in vitro experiments.3 These studies demonstrated as well that both 4- and 4′-OH groups are essential for inducing estrogen receptor down-regulation. DHS was identified for the first time among the metabolites of the trans-stilbene excreted in the urine of guinea pigs, rabbits and mice4 and many years later, the cytochrome P450 isoforms involved in the oxidation of trans-stilbene to DHS were

64 Publications and Patents 5

identified. Shortly after, DHS was found as a phytoalexin in the bark of Yucca periculosa.5 More recently, the effects of DHS on cancer invasion and metastasis were demonstrated in vivo.6

Usually, DHS can be prepared chemically via several syntheses routes including Perkin condensation.7 However, this chemical synthesis involves a sequence of steps, toxic chemicals, and transition metal catalysts with low yield in DHS. With the purpose of meeting the growing demand for stilbenoids, alternative and sustainable approaches for their production are required. The use of enzymes for organic synthesis of fine chemicals and pharmaceuticals has become a powerful alternative to chemical catalysts. Replacing organic/metal catalysts by biocatalysts has several advantages, the most important being excellent regio- and stereo- selectivity, fewer side products, and reduction in environmental impact.

Biocatalysts that preferably catalyze the transfer of an oxygen atom from peroxide to substrates are known as peroxygenases and are classified in a separate sub-subclass, (EC.1.11.2), which was approved in 2011 and at present comprises five members. Among them, the unspecific peroxygenase (UPO, EC 1.11.2.1) is the most prominent one because of its frequency in fungal organisms and promiscuity for oxygen transfer reactions8 that make it highly attractive as industrial biocatalysts.

The first enzyme of this type was discovered in the basidiomycete Agrocybe aegerita.9 UPOs are able to catalyze reactions formerly assigned only to P450s,10 using a similar active site and reaction chemistry.11 However, unlike P450s that are intracellular enzymes and whose activation usually requires an auxiliary enzyme and a source of reducing power, UPOs are secreted proteins, therefore far more 12 stable, and only require H2O2 for activation. Similar enzymes have also been found in other basidiomycetes, such as Coprinellus radians13 and Marasmius rotula14 and there are strong indications for their widespread occurrence in the fungal kingdom.15 Over one-hundred peroxygenase-type genes (encoding enzymes of the heme-thiolate peroxidase superfamily) have been identified during the analysis of 24 basidiomycete genomes16 including Coprinopsis cinerea.17 The C. cinerea peroxygenase has not been isolated to date from the fungus, but one of the peroxygenase genes from its genome was heterologously expressed by Novozymes A/S (Bagsvaerd, Denmark). Very recently, another peroxygenase has been found in the cellulolytic ascomycete Chaetomium globosum.18 These peroxygenases have

65 5 Publications and Patents

been shown to catalyze numerous interesting oxygenation reactions on aromatic compounds19 and later, their ability to oxyfunctionalize diverse aliphatics including linear20–23 and cyclic alkanes as well as complex substrates such as steroids24 and secosteroids25,26 was demonstrated expanding their biotechnological interest.

In this work, the ability of non-recombinant (hereinafter wild) peroxygenases from A. aegerita (AaeUPO), M. rotula (MroUPO) and C. globosum (CglUPO), and recombinant peroxygenase from C. cinerea (rCciUPO), to catalyze the hydroxylation of trans-stilbene is shown. Additionally, the UPO-catalyzed hydroxylation of pinosylvin (Pin, a stilbenoid present in the heartwood of conifers of the family Pinaceae) and RSV was studied.

Experimental

Enzymes

The MroUPO enzyme is a wild peroxygenase isolated from liquid cultures of M. rotula DSM-25031 (a fungus deposited at the German Collection of Microorganisms and Cell Cultures, Braunschweig). MroUPO was purified by fast protein liquid chromatography (FPLC) to apparent homogeneity, and revealed a molecular mass of 32 kDa and an isoelectric point between pH 5.0 and 5.3. The UV-visible spectrum of the enzyme showed a characteristic maximum at 418 nm (Soret band of hemethiolate proteins).14 All media and columns used for enzyme isolation were purchased from GE Healthcare Life Sciences.

The AaeUPO (A. aegerita isoform II, 46 kDa) is another wild peroxygenase, which was isolated from cultures of A. aegerita grown in soybean-peptone medium, and subsequently purified using a combination of Q-Sepharose and SPSepharose and Mono-S ion-exchange chromatographic steps.9

rCciUPO corresponds to the protein model 7249 from the sequenced C. cinerea genome available at the DOE JGI (http://genome.jgi.doe.gov/Copci1). It was expressed in Aspergillus oryzae (patent WO/2008/119780), purified using a combination of S-Sepharose and SP-Sepharose ion-exchange chromatography, and provided by Novozymes A/S (Bagsvaerd, Denmark) as a protein preparation. The recombinant peroxygenase is a glycoprotein with a molecular mass around 44 kDa, a typical UV-vis spectrum with the Soret band at 418 nm, and the ability to

66 Publications and Patents 5

oxygenate different aromatic compounds including veratryl alcohol with a specific activity of approx. 100 U mg−1 (measured as described below).

CglUPO (36 kDa) is a third wild peroxygenase recently isolated from cultures of C. globosum DSM-62110 (German Collection of Microorganisms and Cell cultures), and purified by ammonium sulfate precipitation and successive FPLC using Q-Sepharose FF (ion exchange), Superdex75 (size exclusion), and Mono Q (ion exchange) columns.18 All chromatographic steps were accomplished with an ÄKTA purifier FPLC system (GE Healthcare).

The purified proteins were electrophoretically homogenous, as shown by sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under denaturing conditions, although some minor contaminant proteins were present in CglUPO (ESI† Fig. S1).

One activity unit (U) is defined as the amount of enzyme oxidizing 1 μmol of −1 −1 veratryl alcohol to veratraldehyde (ε310 9300 M cm ) in 1 min at 24 °C, pH 7.0 (the optimum pH for benzylic hydroxylation by AaeUPO and some other peroxygenases) after addition of 2.5 mM H2O2.

Model compounds

trans-Stilbene ([E]-1,2-diphenylethene) and some stilbenoids, including Pin (3,5-dihydroxy-trans-stilbene) and RSV (3,5,4′-trihydroxy-trans-stilbene) were tested as substrates of the above UPOs. Additionally, 4-hydroxy-trans-stilbene (4HS), 4,4′-dihydroxy-trans-stilbene (DHS), oxyresveratrol (oxyRSV, 3,5,2′,4′- tetrahydroxy-trans-stilbene) and trans-stilbene epoxide were used as standards in GC-MS and HPLC analyses. Chemical structures of these compounds are shown in ESI† (Fig. S2).

Enzyme reactions

Enzymatic reactions were performed at 0.1 mM substrate concentration (1 mL reaction volume) and 30 °C, in 50 mM phosphate buffer (pH 7 was selected for comparison). Prior to use, the substrate was dissolved in acetone and added to the buffer to give a final acetone concentration in the reaction of 20% (v/v) (resulting in substrate solubilization and best reaction rate); reactions without acetone were also performed. Ascorbic acid was added to the reaction mixture (except for CglUPO reactions) to prevent further oxidation of hydroxylated aromatic (phenolic)

67 5 Publications and Patents

products and subsequent radical coupling,27 the best results being obtained using 8 mM concentration. Enzyme and H2O2 concentration were optimized for each substrate in the following ranges: AaeUPO (6–500 nM) with 1.2–10 mM H2O2, MroUPO (0.2–1 μM) with 0.5–10 mM H2O2, rCciUPO (0.2–1.2 μM) with 0.5–10 mM H2O2, and CglUPO (0.2–10 μM) with 2.5–10 mM H2O2 (and those resulting in maximal conversion were used). Final H2O2 concentrations were selected for maximal transformation and reaction selectivity. In control experiments, substrates were treated under the same conditions (including H2O2) but without enzyme.

After 30–60 min reaction, products of stilbene conversion were extracted with 18 methyl tert-butyl ether and dried under N2. Enzymatic reactions with O-labeled 18 hydrogen peroxide (H2 O2, 90% isotopic content) from Sigma-Aldrich (2% w: v solution) were performed under the same conditions described above. N,O- Bistrimethylsilyl)trifluoro-acetamide (Supelco) was used to prepare trimethylsilyl (TMS) derivatives of stilbene reaction products that were analyzed by GC-MS. Due to the relatively higher water solubility of RSV and oxyRSV compared with stilbene (which made it difficult to extract them with apolar solvents), products from Pin and RSV conversion were analyzed in a different way. After 30 or 60 min, reactions were stopped with 200 μL of 50 mM sodium azide solution by vigorous shaking and compounds present were analyzed by HPLC. Definitions of some transformation parameters are provided in ESI†.

Enzyme kinetics

The kinetics of stilbene dihydroxylation was studied in two steps. First, stilbene reactions were carried out in 1 mL vials with 3–200 μM substrate; 32 nM MroUPO; 40 nM rCciUPO; 4 nM AaeUPO; 8 mM ascorbic acid; 20% (v/v) acetone. The second hydroxylation step was studied analogously with 3–800 μM of 4-hydroxy- trans-stilbene; 13 nM MroUPO; 120 nM rCciUPO; 4 nM AaeUPO; 8 mM ascorbic acid; 20% (v/v) acetone. The enzyme concentrations were chosen to obtain similar transformation degrees in both reaction steps. The reactions were initiated with 0.5 mM H2O2 and stopped after 10 s by vigorous shaking with 200 μL of 50 mM sodium azide. This short reaction time was selected to: i) obtain maximal turnover numbers; and ii) avoid further oxidation of the first product 4HS enabling the separate analysis of both reactions. All reactions were carried out in triplicate. Product quantification was carried out by GC-MS using external standard curves, and kinetic parameters (kcat, Km) were obtained by fitting the data to the Michaelis–

68 Publications and Patents 5

Menten equation using SigmaPlot software (Systat Softwarwe Inc., San Jose, CA, USA). For total turnover number (TTN) determination of trans-stilbene oxidation, reactions were performed with AaeUPO (13 nM), MroUPO (0.2 μM) and rCciUPO (1.2 μM) and substrate concentration was increased up to 5 mM in the presence of 20% (v/v) of acetone.

GC-MS analyses

The analyses were performed with a Shimadzu GC-MS QP 2010 Ultra system, using a fused-silica DB-5HT capillary column (30 m × 0.25 mm internal diameter, 0.1 μm film thickness) from J&W Scientific. The oven was heated from 50 °C (1.5 min) to 90 °C (2 min) at 30 °C min−1, and then from 90 °C to 250 °C (15 min) at 8 °C min−1. The injection was performed at 250 °C and the transfer line was kept at 300 °C. Compounds were identified by comparing their mass spectra and retention times with those of the authentic standards (4HS, DHS, trans-stilbene epoxide) and quantified from total ion peak areas. Substrate conversion rates and relative molar abundances of products were calculated with relative response factors from chromatographic runs of mixtures of authentic standards. Yields of main products were calculated using standard curves.

HPLC analyses

HPLC analyses were performed with a Shimadzu LC-2030C 3D system equipped with a photo diode array detector using a reversed phase column Agilent InfinityLab Poroshell 120 Bonus-RP C18 (4.6 mm diameter by 150 mm length, 2.7 μm particle size). The column was isocratically eluted at 40 °C and 1 mL min−1 with 0.1% phosphoric acid (pH 2.2) and acetonitrile, 90:10, for 2 min, followed by an 18 min linear gradient to 70% acetonitrile, finally held for 3 min. Oxidation products were identified by comparing their retention times and spectral data with authentic standards (Pin, RSV and OxyRSV). Quantification was obtained from peak areas recorded at 303 nm, and substrate conversion rates and relative abundances of products were calculated as described above.

UPO molecular structures

Both the A. aegerita and the M. rotula UPOs have been crystallized and their solved molecular structures are available at the Protein Data Bank (www.rcsb.org/pdb) with ID 2YP1 and 5FUJ, respectively, among other entries. In

69 5 Publications and Patents

the present study, homology models for the C. cinerea and C. globosum UPOs were obtained at the SWISS-MODEL protein structure homology-modelling server (https://swissmodel.expasy. org).28,29 PyMOL Molecular Graphics System, ver 2.0 Schrödinger, LLC (https://pymol.org) and Swiss-PdbViewer ver 4.1 (https://spdbv.vital-it.ch) were used to examine and illustrate the above UPOs' molecular structures.

Results and discussion

Hydroxylation of stilbene by several UPOs

In the present work, several fungal peroxygenases namely AaeUPO, rCciUPO, MroUPO and CglUPO, were tested for their ability to hydroxylate trans-stilbene using H2O2 as co-substrate (electron acceptor and oxygen source) (Table 1). Complete conversion (100%) was achieved with all UPOs. However, very different regioselectivities were accomplished by the basidiomycetous and ascomycetous UPOs (Fig. 1A). The former (AaeUPO, rCciUPO and MroUPO) hydroxylated selectively the para-positions of the substrate to form 4,4′-dihydroxy-trans-stilbene (DHS) (about 99% of total products), which was identified by comparing GC-MS data (mass spectrum, retention time) with an authentic standard (ESI,† Fig. S3A and S4A-B, respectively). At shorter incubation times the monohydroxylated 4HS was detected (Fig. 1B). Traces of a trihydroxylated-trans-stilbene (THS), most probably formed by hydroxylation of DHS, were also found (Table 1). In contrast, CglUPO did not hydroxylate the aromatic rings and instead formed exclusively the epoxide (trans-stilbene epoxide) in the aliphatic chain of trans-stilbene. trans- Stilbene epoxide was unambiguously identified by GC-MS, comparing its mass spectrum and retention time with an authentic standard (Fig. S5A and S4C-D†, respectively). Product yields (of both DHS and the epoxide) were over 90% for all the UPOs tested. The yield for DHS is much higher than that of chemical synthesis (only 25%) (Scheme S1†). 18O-labeling studies were performed in trans-stilbene 18 reactions using H2 O2 as UPO co-substrate to confirm the source of oxygen incorporated during the formation of both, DHS and the epoxide (Fig. S3B and S5B†). Mass spectral analysis of the resulting dihydroxylated TMS-derivatives (Fig. S3†) showed that the characteristic molecular ion fragments had shifted by approx. 90% from the natural abundance at m/z 356 (found in the reaction with 16 ordinary H2 O2; Fig. S3A†) to m/z 360 (Fig. S3B†; 10% of the original fragments

70 Publications and Patents 5

18 remained in the H2 O2 reactions owing to the 90% isotopic purity of the labeled peroxide). Additionally, in the mass spectrum of the epoxy derivative from the CglUPO reaction, a 90% shift from the natural abundance at m/z 105, m/z 118, m/z + 195 and m/z 196 ([M] ) found in the reaction with unlabeled H2O2 to m/z 107, m/z 120, m/z 197 and m/z 198 was ascertained.

St-epoxide DHS A St

CglUPO rCciUPO MroUPO AaeUPO Control 15 20 25 Retention time (min)

DHS B St

4HS

60 min 15 min 1 min 0 min 15 20 25 Retention time (min)

Fig. 1. GC–MS analysis of trans-stilbene reactions with AaeUPO, rCciUPO, MroUPO and CglUPO and control (without enzyme) showing the remaining substrate (St, trans- stilbene), and the dihydroxylated (DHS, 4,4′-dihydroxy-trans-stilbene) and epoxide (epoxide-St; trans-stilbene oxide) derivatives (A) and GC-MS analysis of stilbene reactions with AaeUPO at different reaction times (0 min, 1 min, 15 min and 60 min) showing the remaining substrate (St), the monohydroxylated (4HS, 4-hydroxy-trans-stilbene) and dihydroxylated (DHS) derivatives (B).

71 5 Publications and Patents

Table 1 Conversion of trans-stilbene (0.1 mM) at 60 min by several peroxygenases and relative molar abundance (%) of products: dihydroxystilbene (DHS), trihydroxystilbene (THS) and stilbene epoxide. The molar yield of DHS (with respect to substrate) is also shown Products (relative molar %) Conversion Main product DHS THS Epoxide (molar %) yield (molar %) AaeUPO (12 nM)a 100 98.6 1.4 - 94 ± 13 rCciUPO (1.2 µM)a 100 98.6 1.4 - 91 ± 7 MroUPO (0.2 µM)b 100 99.1 0.9 - 96 ± 11 CglUPO (3 µM)b 100 - - 100 95 ± 3 a b c d 5 mM H2O2, 10 mM H2O2, DHS, Epoxide.

Kinetics of stilbene oxygenation by UPOs

The estimated kinetic parameters for trans-stilbene hydroxylation by UPOs are summarized in Table 2. AaeUPO was the most efficient enzyme catalyzing the double hydroxylation of trans-stilbene. While the Michaelis–Menten constants (Km) were similar for all enzymes, the turnover numbers (kcat) were much higher for AaeUPO resulting in a 15- and 60-fold increase in the catalytic efficiency (kcat/Km) for the first and second hydroxylation step, respectively. It was also found that 4HS hydroxylation by AaeUPO was more favorable than that of trans-stilbene as it exhibited higher kcat and lower Km values, which increased the catalytic efficiency four times. On the other hand, catalytic efficiencies were roughly in the same range (∼105 M−1 s−1) for both hydroxylation steps in the case of MroUPO and rCciUPO. Compared to most other AaeUPO substrates susceptible to aromatic hydroxylation, the catalytic efficiency for trans-stilbene turned out to be one to two orders of magnitude higher (e.g. for the oxygenation of naphtalene and propranolol, 5 × 105 and 4 × 104 M−1 s−1, respectively, were reported).30,31 This fact can be explained by the higher Km values observed for these substrates reflecting lower affinities to the enzyme compared to trans-stilbene. Similar applies to stilbene hydroxylation by MroUPO, which gave a catalytic efficiency one order of magnitude higher than that reported for naphthalene.14

72 Publications and Patents 5

Table 2 Estimated kinetic parameters for trans-stilbene and 4- hydroxy-trans-stilbene hydroxylation in short (10 s) reactions with several peroxygenases (AaeUPO, MroUPO and rCciUPO). Data represent mean values of three replicates with standard errors -1 -1 -1 kcat (s ) Km (µM) kcat/ Km (M ·s ) St→4HS AaeUPO 85.7 ± 8.5 20.8 ± 5.6 4.1 ± 1.2 x 106 rCciUPO 9.1 ± 0.8 30.5 ± 7.0 3.0 ± 0.7 x 105 MroUPO 4.2 ± 0.2 15.3 ± 2.5 2.7 ± 0.5 x 105

4HS→DHS AaeUPO 225 ± 48.3 14.1 ± 7.7 1.6 ± 0.9 x 107 rCciUPO 6.6 ± 1.3 18.9 ± 8.5 3.5 ± 0.8 x 105 MroUPO 11.7 ± 0.3 46.4 ± 2.3 2.5 ± 0.1 x 105

The catalytic differences between the UPOs oxyfunctionalizing trans-stilbene were supported by their reaction performance. In the case of AaeUPO, TTN for trans-stilbene hydroxylation (982 μM for DHS and 691 μM for 4HS) raised up to 200 000, while using MroUPO and rCciUPO, this number was 8- and 143-fold lower, respectively (Table 3). Thus, AaeUPO displayed a comparably high TTN for the aromatic oxygenation of trans-stilbene as for the benzylic hydroxylation of alkylbenzenes, for which a TTN of 110 000 was achieved.14 Very recently, the synthesis of stilbenoid derivatives by a cytochrome P450 monooxygenase (CYP154E1) from Thermobifida fusca and some of its variants has been reported.32 The enzyme showed a similar para hydroxylating regioselectivity for trans-stilbene and similar conversion rates as reported herein for the UPOs but with considerably lower TTN (4800) as compared with AaeUPO (200 000) and MroUPO (25 000). Comparing kinetic data, CYP154E1 and its variants exhibited lower affinities (higher Km values) and turnover numbers, and hence a thousand-times lower catalytic efficiency compared to AaeUPO.

Table 3 Catalytic performance of several peroxygenases for trans-stilbene hydroxylation TTN TOF (s-1) AaeUPO 200 000 55.6 rCciUPO 1 400 0.4 MroUPO 26 000 7.2

73 5 Publications and Patents

Hydroxylation of stilbene analogs by several UPOs

In addition to DHS synthesis from trans-stilbene, other stilbenoids (trans- stilbene hydroxylated derivatives) such as Pin and RSV were also tested as UPO substrates (Table 4).

Table 4 Conversion of 0.1 mM pinosylvin (Pin) and resveratrol (RSV) by four UPOs yielding RSV, oxyresveratrol (oxyRSV) and oxypinosilvin (oxyPin) (different enzyme doses, reaction times and H2O2 concentrations are compared) Products (relative molar %) Enzyme Substrate Conversion (%) RSV oxyRSV oxyPin AaeUPO 0.025 µM Pina 96.8 62.8 35.2 2.0 0.050 µM Pina 94.0 37.3 62.7 - 0.100 µM Pinb 95.9 21.8 78.2 - 0.144 µM RSVa 71.9 - 100 - rCciUPO 0.1 µM Pinc 80.4 39.6 27.0 33.5 0.1µM RSVb 22.2 - 100 - MroUPO 1 µM Pinc 10.7 100 - - 1 µM RSVb 0 - - - CglUPO 10 µM Pina 0 - - - 10 µM RSVb 0 - - - a b 30 min reactions with 2.5 mM H2O2; 60 min reactions with 5 mM H2O2; c 30 min reactions with 5 mM H2O2

Substrate conversion and product identification was followed/achieved by HPLC (Fig. 2) comparing respective data with authentic standards (Fig. S6†). In Pin reactions, different conversion rates and product patterns were observed in dependence of the particular UPO. While AaeUPO almost completely converted Pin (up to 97%) and rCciUPO did it to large extent (80%), just a low conversion rate (11%) was obtained with MroUPO and no conversion was observed with CglUPO. The conversion of Pin by AaeUPO and rCciUPO yielded RSV that was in turn oxidized to oxyRSV (up to 78% of oxidized product when the highest AaeUPO dose was used) but, in the case of rCciUPO, a significant amount of

74 Publications and Patents 5

another monohydroxylated compound tentatively assigned as 2′-hydroxypinosylvin (oxypinosylvin, oxyPin) was detected (being very minor in the AaeUPO reactions). This suggest that AaeUPO is more selective towards the position 4′ of Pin yielding RSV, while similar amounts of RSV and oxyPin are formed by rCciUPO hydroxylation of Pin at the 4′ and 2′ positions, respectively. With MroUPO, RSV was the only product obtained. When using RSV as substrate, up to 72% and 22% conversion was attained with AaeUPO and rCciUPO, respectively, but no conversion was observed with CglUPO and MroUPO. The reaction of RSV with the two former enzymes generated oxyRSV.

RSV oxyRSV

UPO RSV oxyRSV Aae Pin

5 10 15 20 5 10 15 20 Retention time (min) Retention time (min)

RSV RSV oxyPin

UPO Pin Cci

r oxyRSV oxyRSV

5 10 15 20 5 10 15 20 Retention time (min) Retention time (min)

Pin RSV

UPO Mro

RSV

5 10 15 20 5 10 15 20 Retention time (min) Retention time (min)

Pin RSV

UPO Cgl

5 10 15 20 5 10 15 20 Retention time (min) Retention time (min) Figure 2. HPLC analysis of pinosylvin (Pin, left) and resveratrol (RSV, right) reactions with AaeUPO, rCciUPO, MroUPO and CglUPO showing the remaining substrates (Pin, RSV) and the products, resveratrol (RSV), oxyresveratrol (oxyRSV) and the dihydroxylated derivative of Pin (oxyPin).

75 5 Publications and Patents

Molecular architecture vs. UPO activity

The above differences in the regioselectivity and kinetic constants for stilbene oxidation (Tables 1 and 2) should be relatable to differences in the UPOs' molecular structure affecting the specific residues putatively involved in catalysis. These would be located at: i) the access channel leading to the buried heme cofactor and its surface opening; and ii) the enzyme active site itself, i.e. the heme pocket that follows the access channel.

Besides the obvious difference in molecular size – ‘long’ AaeUPO and rCciUPO (Fig. 3 left) consist of 325 and 337 amino acids (mature protein), respectively, while ‘short’ MroUPO and CglUPO (Fig. 3 right) only comprise 236 and 242 residues, respectively8 – the most pronounced difference is the distribution of solvent- exposed phenylalanines. AaeUPO possesses four phenylalanine residues (Phe69, Phe191, Phe199 and Phe274) at the entrance of the heme access channel, which make it rather hydrophobic and affine to aromatic rings, while MroUPO and rCciUPO only have one (Fig. 3). It is conceivable that aromatic interactions between the substrate rings and these four phenylalanine residues may contribute to optimal positioning of stilbene (and later of 4HS) with the para-position at reaction distance of the ferryl oxygen (Fe4+=O) of activated heme (UPO compound I). The same would apply to other stilbenoids investigated in the present study. Less ‘perfect’ positioning of the substrate in the more aliphatic environment of rCciUPO and MroUPO channels would result in lower hydroxylating efficiency both for stilbene (14–15 times lower) and 4HS (45–65 times lower) (Table 2). For MroUPO, the catalytic efficiencies were similar to those reported for other UPO substrates like veratryl alcohol or benzyl alcohol,14 but the catalytic efficiencies of AaeUPO for the oxyfunctionalization of stilbene and especially of 4HS turned out to be much better than the values reported for any other ‘two-electron oxidation’ substrate.9 The oxygenation results also show that UPOs hydroxylate the phenyl moiety of stilbenoids with preference respect to the 4-OH–phenyl moiety. This is revealed by the production of DHS in the stilbene reaction. The complete reaction most probably involves 4HS release from the enzyme and re-entering with the second aromatic ring pointing towards the heme cofactor.

76 Publications and Patents 5

A B

F323 F310 F312 F191 F69 F274 F170 F160 F199 F46

F155

C D F331

F283 F62 F321 F191 F58 F154 F280 F46

F209

F210

Figure 3. Solvent access surface in AaeUPO (A), MroUPO (B), rCciUPO (C) and CglUPO (D) showing different: i) sizes of the channel giving access to the heme cofactor (CPK- colored sticks); and ii) phenylalanine residues (magenta surface) at the channel edge and other surface regions. Cenital views showing the channels giving access to the distal side of the cofactor pocket on the heme iron atom (one acetate molecule is shown inside the MroUPO channel). From two crystallographic, 2YP1 (A) and 5FJU (C), and two homology models (B and D) corresponding to the secreted mature proteins (therefore, residue numbering does not include the signal peptide).

The regioselectivity of CglUPO in stilbene oxygenation radically differed from that of aromatic oxygenation discussed above, since ethenyl epoxidation was produced. Interestingly, the UPO of the ascomycete C. globosum (A. aegerita, M. rotula and C. cinerea are basidiomycetes), has a noticeably different active site. Of the five residues above the heme plane (Fig. 4), it only shares with AaeUPO the glutamic residue (CglUPO Glu158) putatively involved in the binding and fission of H2O2 (ref. 8) (also conserved in the other two UPOs). An adjacent residue is also thought to contribute to this reaction (as charge stabilizer), being either an arginine

77 5 Publications and Patents

(in AaeUPO and CciUPO; Arg189) or a histidine (in MroUPO and CglUPO; His86 and His88, respectively). CglUPO also shares with MroUPO the absence of a phenylalanine near the heme edge (homologous to AaeUPO Phe121). However, the main peculiarities of CglUPO's active site are: i) the absence of a conserved phenylalanine (homologous to AaeUPO Phe199, although Phe154 occupies a neighbor position) and in particular, ii) the presence of a unique and potentially reactive tyrosine (Tyr162) in the middle of the active site. Whether this tyrosine is responsible for the preferred epoxidation of isolated double bonds within complex molecules – in contrast to other UPOs, CglUPO also specifically epoxidized testosterone18 – will have to be clarified in future mutagenesis studies. Although a yeast expression system for AaeUPO has been developed after enzyme directed evolution,33 new procedures and hosts for expressing wild-type genes of this and other UPOs are required for investigating active-site residues and engineering UPO biocatalysts for selective oxygenation reactions of biotechnological interest.34

A B

T192 F199 I153 Ac H86 R189 Ac F160 E196 E157

I84 F121 C36 C17

C D F154 T192 F199

R189 H88 Y162 E196 E158

L86 F121 C36 C17

Figure 4. Active site residues in AaeUPO (A), MroUPO (B), rCciUPO (C) and CglUPO (D): Five residues above the heme cofactor (heme pocket "distal" side) plus the cysteine acting as the fifth ligand of the heme iron located below the cofactor (heme pocket "proximal" side) are shown, together with an acetate molecule (Ac) occupying the substrate position in the AaeUPO and MroUPO crystals (all as CPK-colored sticks). Black labels correspond to residues as found in AaeUPO, while varying residues have color labels (blue, red or green). See Fig. 3 for model references and other informations.

78 Publications and Patents 5

Conclusions

We report here on a new enzymatic route for the production of DHS from trans- stilbene and RSV from Pin (Scheme 1) in one-step by fungal peroxygenases (UPOs). The potent effect of DHS in inhibition of cancer invasion and metastasis has recently been reported “in vivo”.6 Usually, DHS is obtained by chemical synthesis, which involves several-steps, the use of metal catalysts, etc. (Scheme S1†). The enzymatic reaction with UPOs represents a smart and environmentally sound alternative to chemical synthesis. In addition, the UPO catalyzed synthesis of the potent antioxidants and free radical scavengers RSV (and oxyRSV) from Pin (a natural constituent of pine wood) is also reported here for the first time. A preliminary analysis of the active site cavities and access channels has revealed differences that may explain and affect the reactivity and regioselectivity of the different UPOs in the stilbene oxygenation, but more structural and functional information required for reaction understanding and enzyme engineering.

Scheme 1. Enzymatic synthesis of DHS from trans-stilbene (St) and RSV from Pin.

Conflicts of interest

There are no conflicts to declare.

79 5 Publications and Patents

Acknowledgements

This study was funded by the EnzOx2 (H2020-BBI-PPP-2015-2-1-720297) EU-project, and the BIORENZYMERY (AGL2014-53730-R) project of the Spanish MINECO (co-financed by FEDER). H. Lund (Novozymes) is acknowledged for rCciUPO.

References

1. S. S. Kulkarni and C. Canto, Biochim. Biophys. Acta, Mol. Basis Dis., 2015, 1852, 1114–1123. 2. L. A. Stivala, M. Savio, F. Carafoli, P. Perucca, L. Bianchi, G. Maga, L. Forti, U. M. Pagnoni, A. Albini, E. Prosperi and V. Vannini, J. Biol. Chem., 2001, 276, 22586–22594. 3. M. Savio, T. Coppa, L. Bianchi, V. Vannini, G. Maga, L. Forti, O. Cazzalini, M. C. Lazze, P. Perucca, E. Prosperi and L. A. Stivala, Int. J. Biochem. Cell Biol., 2009, 41, 2493–2502. 4. J. E. Sinsheimer and R. V. Smith, Biochem. J., 1969, 111, 35–41. 5. P. Torres, J. G. Avila, A. R. de Vivar, A. M. Garcia, J. C. Marin, E. Aranda and C. L. Cespedes, Phytochemistry, 2003, 64, 463–473. 6. M. Savio, D. Ferraro, C. Maccario, R. Vaccarone, L. D. Jensen, F. Corana, B. Mannucci, L. Bianchi, Y. H. Cao and L. A. Stivala, Sci. Rep., 2016, 6, 19973. 7. G. A. Locatelli, M. Savio, L. Forti, I. Shevelev, K. Ramadan, L. A. Stivala, V. Vannini, U. Hubscher, S. Spadari and G. Maga, Biochem. J., 2005, 389, 259– 268. 8. M. Hofrichter, H. Kellner, M. J. Pecyna and R. Ullrich, Adv. Exp. Med. Biol., 2015, 851, 341–368. 9. R. Ullrich, J. Nuske, K. Scheibner, J. Spantzel and M. Hofrichter, Appl. Environ. Microbiol., 2004, 70, 4575–4581. 10. R. Bernhardt, J. Biotechnol., 2006, 124, 128–145. 11. K. Piontek, E. Strittmatter, R. Ullrich, G. Gröbe, M. J. Pecyna, M. Kluge, K. Scheibner, M. Hofrichter and D. A. Plattner, J. Biol. Chem., 2013, 288, 34767– 34776. 12. M. Hofrichter and R. Ullrich, Curr. Opin. Chem. Biol., 2014, 19, 116–125. 13. D. H. Anh, R. Ullrich, D. Benndorf, A. Svatos, A. Muck and M. Hofrichter, Appl. Environ. Microbiol., 2007, 73, 5477–5485.

80 Publications and Patents 5

14. G. Gröbe, M. Ullrich, M. Pecyna, D. Kapturska, S. Friedrich, M. Hofrichter and K. Scheibner, AMB Express, 2011, 1, 31–42. 15. M. J. Pecyna, R. Ullrich, B. Bittner, A. Clemens, K. Scheibner, R. Schubert and M. Hofrichter, Appl. Microbiol. Biotechnol., 2009, 84, 885–897. 16. D. Floudas, M. Binder, R. Riley, K. Barry, R. A. Blanchette, B. Henrissat, A. T. Martínez, R. Otillar, J. W. Spatafora, J. S. Yadav, A. Aerts, I. Benoit, A. Boyd, A. Carlson, A. Copeland, P. M. Coutinho, R. P. de Vries, P. Ferreira, K. Findley, B. Foster, J. Gaskell, D. Glotzer, P. Górecki, J. Heitman, C. Hesse, C. Hori, K. Igarashi, J. A. Jurgens, N. Kallen, P. Kersten, A. Kohler, U. Kües, T. K. A. Kumar, A. Kuo, K. LaButti, L. F. Larrondo, E. Lindquist, A. Ling, V. Lombard, S. Lucas, T. Lundell, R. Martin, D. J. McLaughlin, I. Morgenstern, E. Morin, C. Murat, M. Nolan, R. A. Ohm, A. Patyshakuliyeva, A. Rokas, F. J. Ruiz-Dueñas, G. Sabat, A. Salamov, M. Samejima, J. Schmutz, J. C. Slot, F. St. John, J. Stenlid, H. Sun, S. Sun, K. Syed, A. Tsang, A. Wiebenga, D. Young, A. Pisabarro, D. C. Eastwood, F. Martin, D. Cullen, I. V. Grigoriev and D. S. Hibbett, Science, 2012, 336, 1715–1719. 17. J. E. Stajich, S. K. Wilke, D. Ahren, C. H. Au, B. W. Birren, M. Borodovsky, C. Burns, B. Canbäck, L. A. Casselton, C. K. Cheng, J. X. Deng, F. S. Dietrich, D. C. Fargo, M. L. Farman, A. C. Gathman, J. Goldberg, R. Guigo, P. J. Hoegger, J. B. Hooker, A. Huggins, T. Y. James, T. Kamada, S. Kilaru, C. Kodira, U. Kües, D. Kupfert, H. S. Kwan, A. Lomsadze, W. X. Li, W. W. Lilly, L. J. Ma, A. J. Mackey, G. Manning, F. Martin, H. Muraguchi, D. O. Natvig, H. Palmerini, M. A. Ramesh, C. J. Rehmeyer, B. A. Roe, N. Shenoy, M. Stanke, V. Ter Hovhannisyan, A. Tunlid, R. Velagapudi, T. J. Vision, Q. D. Zeng, M. E. Zolan and P. J. Pukkila, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 11889–11894. 18. J. Kiebist, K. U. Schmidtke, J. Zimmermann, H. Kellner, N. Jehmlich, R. Ullrich, D. Zänder, M. Hofrichter and K. Scheibner, ChemBioChem, 2017, 18, 563–569. 19. M. Hofrichter, R. Ullrich, M. J. Pecyna, C. Liers and T. Lundell, Appl. Microbiol. Biotechnol., 2010, 87, 871–897. 20. E. D. Babot, J. C. del Río, L. Kalum, A. T. Martínez and A. Gutiérrez, Biotechnol. Bioeng., 2013, 110, 2332. 21. A. Olmedo, C. Aranda, J. C. del Río, J. Kiebist, K. Scheibner, A. T. Martínez and A. Gutiérrez, Angew. Chem., Int. Ed., 2016, 55, 12248–12251.

81 5 Publications and Patents

22. A. Gutiérrez, E. D. Babot, R. Ullrich, M. Hofrichter, A. T. Martínez and J. C. del Río, Arch. Biochem. Biophys., 2011, 514, 33–43. 23. S. Peter, M. Kinne, X. Wang, R. Ulrich, G. Kayser, J. T. Groves and M. Hofrichter, Rev. Geophys., 2011, 278, 3667–3675. 24. E. D. Babot, J. C. del Río, M. Cañellas, F. Sancho, F. Lucas, V. Guallar, L. Kalum, H. Lund, G. Gröbe, K. Scheibner, R. Ullrich, M. Hofrichter, A. T. Martínez and A. Gutiérrez, Appl. Environ. Microbiol., 2015, 81, 4130–4142. 25. E. D. Babot, J. C. del Río, L. Kalum, A. T. Martínez and A. Gutiérrez, ChemCatChem, 2015, 7, 283–290. 26. F. Lucas, E. D. Babot, J. C. del Río, L. Kalum, R. Ullrich, M. Hofrichter, V. Guallar, A. T. Martínez and A. Gutiérrez, Catal. Sci. Technol., 2016, 6, 288– 295. 27. M. Kinne, M. Poraj-Kobielska, E. Aranda, R. Ullrich, K. E. Hammel, K. Scheibner and M. Hofrichter, Bioorg. Med. Chem. Lett., 2009, 19, 3085–3087. 28. M. Biasini, S. Bienert, A. Waterhouse, K. Arnold, G. Studer, T. Schmidt, F. Kiefer, T. G. Cassarino, M. Bertoni, L. Bordoli and T. Schwede, Nucleic Acids Res., 2014, 42, W252–W258. 29. N. Guex, M. C. Peitsch and T. Schwede, Electrophoresis, 2009, 30, S162– S173. 30. M. Poraj-Kobielska, J. Atzrodt, W. Holla, M. Sandvoss, G. Gröbe, K. Scheibner and M. Hofrichter, J. Labelled Compd. Radiopharm., 2013, 56, 513– 519. 31. M. G. Kluge, R. Ullrich, K. Scheibner and M. Hofrichter, Appl. Microbiol. Biotechnol., 2007, 75, 1473–1478. 32. A. Rühlmann, D. Antovic, T. J. J. Müller and V. B. Urlacher, Adv. Synth. Catal., 2017, 359, 984–994. 33. P. Molina-Espeja, S. Ma, D. M. Maté, R. Ludwig and M. Alcalde, Enzyme Microb. Technol., 2015, 73–74, 29–33. 34. Y. Wang, D. Lan, R. Durrani and F. Hollmann, Curr. Opin. Chem. Biol., 2017, 37, 1–9.

82 Publications and Patents 5

Supporting information

Selective synthesis of the resveratrol analogue 4,4′-dihydroxy- trans-stilbene and stilbenoids modification by fungal peroxygenases

Carmen Aranda,a René Ullrich,b Jan Kiebist,c Katrin Scheibner,c José C. del Río,a Martin Hofrichter,b Angel T. Martínez,d and Ana Gutiérrez *a

a Instituto de Recursos Naturales y Agrobiología de Sevilla, CSIC, Reina Mercedes 10, E-41012, Seville, Spain. E-mail: [email protected] b TU Dresden, Department of Bio- and Environmental Sciences, Markt 23, 02763 Zittau, Germany c JenaBios GmbH, Löbstedter Str. 80, 07749 Jena, Germany d Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu 9, E-28040 Madrid, Spain

This Supporting Information includes Supporting Methods, Figures S1-S6 and Scheme S1.

83 5 Publications and Patents

Supporting Methods (definitions of process parameters):

- Conversion was calculated deducting the remaining substrate from the total concentration of substrate at the beginning of the reaction.

- Selectivity is referred to percentage (%) of one product within the total amount of products.

- Turnover number (kcat) is the maximum number of chemical conversions of substrate molecules per second that a single catalytic site will execute for a given enzyme concentration. It was obtained by fitting the data to the Michaelis-Menten equation.

- Total turnover number (TTN) was calculated dividing the mol of products (multiplied by the number of transformation suffered compared to substrate) by the mol of enzyme in the reaction.

- Yield was estimated comparing the mol of product obtained (calculated by GCMS with external standard curves) with the initial mol of substrate in the reaction.

84 Publications and Patents 5

Figure S1. SDS PAGE of different UPO preparations, from left to right: rCciUPO, AaeUPO, MroUPO and CglUPO. 10-12% Bis-Tris was used, and the proteins were visualized with a colloidal Blue staining (Invitrogen). Conditions (50 mM dithiothreitol) resulted in monomeric MroUPO. Low molecular weight standards (Thermo Scientific, Darmstadt, Germany) were included.

85 5 Publications and Patents

Figure S2. Chemical structures of trans-stilbene (St); 4-hydroxy-trans-stilbene (4HS); 4,4′-dihydroxy-trans-stilbene (DHS); trans-stilbene epoxide (St-epoxide); pinosylvin (Pin), resveratrol (RSV), oxypinosylvin (oxyPin) and oxyresveratrol (oxyRSV).

86 Publications and Patents 5

16OTMS A 356 [M+] 100% TMS16O

73

[M-CH ]+ 163 3 341 59 133 235 267 307 372 50 100 150 200 250 300 350 400 m/z

18OTMS B 360 [M+] 100% TMS18O

73

+ [M-CH3] 165 345 59 137 237 269 311 370 50 100 150 200 250 300 350 400 m/z

Figure S3. Mass spectra of 4,4′-dihydroxy-trans-stilbene from enzymatic reactions of AaeUPO, rCciUPO and MroUPO with trans-stilbene in O-labeling experiments (B) and controls (A). The formulae for the unlabeled compounds (A) and the labeled compounds (B) are shown as trimethylsilyl (TMS) derivatives.

87 5 Publications and Patents

A DHS C St-epoxide

15 20 25 15 20 25 Retention time (min) Retention time (min) B D DHS St-epoxide

15 20 25 15 20 25 Retention time (min) Retention time (min)

Figure S4. GC-MS chromatograms of 4,4′-dihydroxy-trans-stilbene (DHS) from trans-stilbene reaction with AaeUPO (A) and DHS standard (B); and of trans- stilbene epoxide (St-epoxide) from trans-stilbene reaction with CglUPO (C) and St-epoxide standard (D).

88 Publications and Patents 5

16O A

100% 167

89

195 196 [M]+ 77 165 178 63 105 152 51 83 98 118 50 100 150 200 m/z

18O B

167 100% 89

165 178 197 + 77 198 [M] 63 107 152 51 83 120

50 100 150 200 m/z

Figure S5. Mass spectra of trans-stilbene epoxide from enzymatic reactions of CglUPO with trans-stilbene in O-labeling experiments (B) and controls (A). The formulae for the unlabeled compounds (A) and the labeled compounds (B) are shown.

89 5 Publications and Patents

A RSV

oxyRSV Pin

5 10 15 20 Retention time (min)

Pin B

RSV oxyRSV

5 10 15 20 Retention time (min)

Figure S6. HPLC chromatograms of pinosylvin (Pin), resveratrol (RSV) and oxyRSV from Pin reaction with AaeUPO (A) and a mixture of authentic standards (0.1 mM each) of Pin, RSV and oxyRSV (B).

90 Publications and Patents 5

Scheme S1. Comparison of step number and yield in (a) chemical,7 and (b) enzymatic using UPO (this study) synthesis of 4,4′-dihydroxy-trans-stilbene (DHS).

a) Chemical synthesis of DHS OAc

a) N(CH CH ) , HO 2 3 3 O O O O O c) MeOH, 180 ºC (N ,6 h) OH K2CO3 + 2 HO O b) HCl 5%, HO CH3CH2Ac, OH 25 C OH OAc

O d) Cu₂Cr₂O₅, OH 240 ºC (N2 ,5 h) e) HCl 5%, CH CH Ac HO 3 2

OH Yield: 25% b) Enzymatic synthesis of DHS

UPO

H2O2 Yield: 94%

91

Publications and Patents 5

5.1. Publication II: Aranda C., Municoy M., Guallar V., Kiebist J., Scheibner K., Ullrich R., del Río J.C., Hofrichter M., Martínez A.T. and Gutiérrez A. (2019). Selective synthesis of 4-hydroxyisophorone and 4-ketoisophorone by fungal peroxygenases. Catalysis Science and Technology (Published online). DOI: 10.1039/C8CY02114G.

93

Publications and Patents 5

Selective Synthesis of 4-Hydroxyisophorone and 4- Ketoisophorone by Fungal Peroxygenases

Carmen Aranda,a Martí Municoy,b Víctor Guallar,b,c Jan Kiebist,d Katrin Scheibner,d René Ullrich,e José C. del Río,a Martin Hofrichter,e Angel T. Martínez,*f and Ana Gutiérrez *a

a Instituto de Recursos Naturales y Agrobiología de Sevilla, CSIC, Reina Mercedes 10, E-41012 Seville, Spain. E-mail: [email protected] b Barcelona Supercomputing Center, Jordi Girona 31, E-08034, Barcelona, Spain c ICREA Passeig Lluís Companys 23, E-08010, Barcelona, Spain d JenaBios GmbH, Löbstedter Str. 80, 07749 Jena, Germany e TU Dresden, Department of Bio- and Environmental Sciences, Markt 23, 02763 Zittau, Germany d Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu 9, E-28040 Madrid, Spain, E-mail: [email protected]

* Corresponding author

Abstract: The recently discovered unspecific peroxygenases (UPO) from the ascomycetes Chaetomium globosum and Humicola insolens were capable of selectively hydroxylating isophorone to 4-hydroxyisophorone (4HIP) and 4-ketoisophorone (4KIP), which are substrates of interest for the pharmaceutical and flavor-and- fragrance sectors. The model UPO from the basidiomycete Agrocybe aegerita was less regioselective, forming 7-hydroxyisophorone (and 7-formylisophorone) in addition to 4HIP. However, it was the most stereoselective UPO yielding the S- enantiomer of 4HIP with 88% ee. Moreover, using H. insolens UPO full kinetic resolution of racemic HIP was obtained within only 15 min, with >75% recovery

95 5 Publications and Patents

of the R-enantiomer. Surprisingly, the UPOs from two other basidiomycetes, Marasmius rotula and Coprinopsis cinerea, failed to transform isophorone. The different UPO selectivities were rationalized by computational simulations, in which isophorone and 4HIP were diffused into the enzymes using the adaptive PELE software, and the distances from home bound oxygen in H2O2-activated enzyme to different substrate atoms, and the binding energies were analyzed. Interestingly, for process upscaling, full conversion of 10 mM isophorone was achieved with H. insolens UPO within nine hours, with total turnover numbers up to 5500. These biocatalysts, which only require H2O2 for activation, may represent a novel, simple and environmentally-friendly route for the production of isophorone derivatives.

Introduction

Isophorone derivatives, such as 4-hydroxyisophorone (4HIP) and 4- ketoisophorone (4KIP), are of interest as flavour-and-fragrance additives,1 and as intermediates in the synthesis of pharmaceuticals, vitamins and natural pigments.2,3 A variety of chemical methods is available for the production of 4HIP and 4KIP. Thus, both derivatives have been synthesized from β-isophorone,1,4 which – on its part – can be obtained by isomerization of isophorone (also known as α- isophorone). The rearrangement of β-isophorone to the α-isomer, however, is a main drawback of this process. The direct oxidation of isophorone to 4KIP with molecular oxygen (O2) appeared to be the solution, using ‘copper (II) chloride- acetylacetone’ or molybdenum-based systems as catalysts,5 but required toxic heavy metals and led to the formation of undesired side products. Moreover, a direct chemical oxidation process of isophorone to 4HIP is not available and this compound is usually synthesized by reduction of 4KIP,6,7 which can be a rather expensive starting material.

Alternatively, some biological processes for the synthesis of 4HIP and 4KIP have been described, often using cytochrome P450 monooxygenases (P450s). Among them, the microbial biotransformation of isophorone was described for fungi – like Aspergillus niger, Alternaria alternata and Neurospora crassa - 8,9 with 4HIP and 7-hydroxyisophorone (7HIP) as main metabolites. More recently, a process using recombinant Escherichia coli transformed with the P450-BM3 gene (together with the gene of NADPH-regenerating glucose dehydrogenase) allowed the scaled-up selective production of 4HIP at kilogram scale.10 On the other hand,

96 Publications and Patents 5

4KIP has been produced either in an one-pot two-step enzymatic process or as a cascade process employing cells co-expressing P450-WAL and Cm-ADH10 dehydrogenase.11 However, isolated P450s generally suffer the disadvantage of rather higher instability and the frequent need of auxiliary enzymes/domains and expensive cofactors.

Unspecific peroxygenases (UPOs, EC.1.11.2.1) are novel and appealing biocatalysts for organic synthesis, since their ‘simplicity’ (only H2O2 is required for activation) and stability (as secreted enzymes) circumvent major disadvantages of P450s while catalyzing the same kind of oxyfunctionalization reactions.12 The first enzyme of this class was discovered in 2004 in the basidiomycete Agrocybe aegerita13 and since then, new peroxygenases came out from Coprinellus radians14, Marasmius rotula,15 and more recently from Chaetomium globosum.16 Their widespread occurrence in the fungal kingdom has been demonstrated by the analysis of basidiomycete, ascomycete and other fungal genomes and revealed over one-thousand putative peroxygenase genes.17 This allowed the production of recombinant enzymes, like those of Coprinopsis cinerea18 or Humicola insolens,16 which are heterologously expressed by Novozymes A/S (Bagsvaerd, Denmark) in the mold Aspergillus oryzae.

The spectrum of reactions catalyzed by these enzymes is steadily increasing and includes oxygenations of both aromatic19,20 and aliphatic compounds,21-24 fatty acids epoxidation25 and chain-shortening,26 and also reactions of rather complex and bulky substrates like steroids,27,28 and secosteroids29,30 that are subject to epoxidation, side-chain hydroxylation or side-chain removal.

In the present work, the hydroxylation of isophorone by several UPOs with different selectivities is presented for the first time, to be included in the portfolio of reactions catalyzed by these novel and exciting enzymes.12,17

Experimental

Enzymes

AaeUPO (isoform II, 46 kDa), the first UPO described in 2004, is a wild-type (i.e. non-recombinant) peroxygenase from cultures of A. aegerita TM-A1, grown in soybean-peptone medium, which was purified as described by Ullrich and

97 5 Publications and Patents

Hofrichter.31 MroUPO is another wild-type peroxygenase (32 kDa) from cultures of M. rotula DSM-25031 (German Collection of Microorganisms and Cell Cultures, Braunschweig), which was purified as described by Gröbe et al.15

CglUPO (36 kDa) is a third wild-type peroxygenase from cultures of C. globosum DSM-62110, which was purified as recently described by Kiebist et al.16 The recombinant enzymes rCciUPO (44 kDa) and rHinUPO were provided by Novozymes A/S. rCciUPO corresponds to the protein model 7249 from the sequenced C. cinerea genome available at the JGI (http://genome.jgi.doe.gov/Copci1) used in several studies.22,27,30 The rHinUPO sequence has been more recently reported32 and used for oxyfunctionalizations that are not catalyzed by other UPOs.16 Both UPOs were expressed by Novozymes in Aspergillus oryzae (patent WO/2008/119780). All UPO proteins were purified by fast protein liquid chromatography (FPLC) using a combination of size exclusion chromatography (SEC) and ion exchange chromatography on different anion and cation exchangers. Purification was confirmed by sodium dodecylsulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and UV-visible spectroscopy following the characteristic heme-maximum around 420 nm (Soret band of resting-state heme-thiolate proteins).

Enzyme concentration was estimated according to the characteristic UV-Vis band of the reduced UPO-complex with carbon monoxide.33

Model compounds

3,5,5-Trimethyl-2-cyclohexen-1-one (isophorone; also known as α-isophorone) from Sigma Aldrich (97% purity) was tested as substrate of the above UPOs. 3,5,5- Trimethyl-2-cyclohexen-1,4-dione (4-ketoisophorone, 4KIP) also from Sigma Aldrich and chemically-synthesized 4-hydroxy-3,5,5-trimethyl-2-cyclohexen-1- one (4-hydroxyisophorone, 4HIP) by 4KIP reduction, were used as standards in gas chromatography-mass spectrometry (GC-MS) analyses. 4HIP, obtained as a racemic mixture by chemical reduction of 4KIP,7 was used as substrate in enzymatic reactions together with isophorone.

Enzyme reactions

Reactions (1-mL volume) with isophorone (0.1 mM) were performed at 30 °C, in 50 mM phosphate buffer, pH 7. The enzyme concentrations ranged from 50 nM

98 Publications and Patents 5

to 10 μM, using 2.5 to 5 mM H2O2. In control experiments, substrates were treated under the same conditions (including H2O2) but without enzyme. After 30-min reaction, products were extracted with methyl tert-butyl ether, which was evaporated under N2 and the products dissolved in chloroform to be analyzed by GC-MS. Reactions at higher isophorone concentration (10 mM) were performed with CglUPO and rHinUPO (2-5 μM). H2O2 was added with a syringe pump to give concentrations in the reaction mixture of 1 or 5 mM h-1 during 48 h or 12 h, respectively. Reactions for chiral analyses were carried out with 10 mM isophorone or 4HIP (racemic mixture) and 5 μM enzyme (CglUPO, rHinUPO and AaeUPO) for 60 min, with H2O2 manually added in small doses to a final concentration of 5- 20 mM. Products were extracted with ethyl acetate and directly analyzed by GC- MS or dried and dissolved in the mobile phase to be analyzed by HPLC.

Enzyme kinetics

Reactions were carried out with 6.25-6400 μM substrate and 100 nM enzyme. They were initiated adding 0.5 mM H2O2 and stopped by vigorous shaking in 5 mM sodium azide. Reaction times, the reaction velocity of which was linear, were previously selected: 5 min for AaeUPO, 3 min for CglUPO and 1 min for rHinUPO. All reactions were performed in triplicates. Product quantification was carried out by GC-MS using external standard curves, and kinetic parameters - turnover number (kcat), Michaelis constant (Km) and catalytic efficiency (kcat/Km) - were obtained by fitting the data to the Michaelis-Menten equation, or to the corresponding variation of this equation when substrate inhibition is occurring, using SigmaPlot (Systat Softwarwe Inc., San Jose, CA, USA).

GC-MS

The analyses were performed in a Shimadzu GC-MS QP 2010 Ultra system, using a fused-silica DB-5HT capillary column (30 m × 0.25 mm internal diameter, 0.1 μm film thickness) from J&W Scientific. The oven was heated from 50 °C (1.5 min) to 90 °C (2 min) at 30 °C min−1, and then from 90 °C to 250 °C (15 min) at 8 °C min−1. The injection was performed at 250 °C and the transfer line was kept at 300 °C. Compounds were identified by comparing their mass spectra and retention times with those of available commercial or synthesized authentic standards, and by search in the NIST library.

99 5 Publications and Patents

Chiral HPLC

Chiral analyses were performed with a Shimadzu LC-2030C 3D system equipped with a photo-diode array detector using a chiral column Chiralpak IG (5 μM particle size, 4.6 mm diameter x 150 mm, Daicel Chemical Industries Ltd.) equipped with a Chiralpak IG guard column (5 μM particle size, 4.0 mm diameter x 10 mm). The column was eluted in isocratic mode with 95% hexane and 5% isopropanol at 0.5 mL min-1 for 60 min and the absorbance was monitored at 254 nm. Enantiomers were identified based on the elution order previously reported.34

Molecular models

AaeUPO and MroUPO models were obtained from the chain-A of the 2YP1 and 5FUJ crystal structures, after removing the Cys227 dimerization disulfide bridge from the second structure. Homology models for CglUPO, rHinUPO and rCciUPO were obtained at the Swiss-Model server,35,36 with related crystal structures as templates. The heme cofactor and the Mg ion (along with its two coordinated water molecules) were then superimposed using 2YP1 and 5FUJ as templates, followed by an initial minimization, to release steric clashes. An oxygen atom (iron-oxo) was finally added to all structures modeling heme compound I. All systems were prepared at pH 7 with the protein preparation wizard from Schrödinger.37 Heme charges were obtained from a quantum mechanics/molecular mechanics (QM/MM) minimization using QSite at the DFT M06-L(lacvp*)/OPLS level of theory. Based on previous experience in heme-bound systems,38 the charge of the Mg ion was set to 1.2. isophorone and 4HIP (S- and R-enantiomers) substrates were built with Maestro and optimized at the OPLS level of theory. In addition, two explicit water molecules were placed in the active site when exploring diffusion of the ligands. The presence of a water molecule has been highlighted in compound I activity and might be important when diffusing polar substrates such as 4HIP.39

Ligand diffusion simulations

The new adaptive-PELE software40 was used to study ligand diffusion and binding on the different UPO structures. PELE uses a Monte Carlo (MC) procedure to describe the protein-ligand conformational dynamics. At each MC iteration, the algorithm performs: 1) ligand perturbation (translation and rotation); 2) protein perturbation following normal modes; 3) explicit water sampling; 4) side chain prediction; and 5) overall minimization. The final structure is then accepted or

100 Publications and Patents 5

rejected based on a Metropolis criterion. The adaptive protocol improves PELE’s sampling by running multiple short simulations (epochs) where initial conditions are selected through a reward function aiming at sampling non visited areas. We also used a new version of PELE that allows for explicit water sampling, where water molecules are allowed to freely move (with 100 small translations and rotations) after the backbone sampling. Two sets of simulations were performed for each ligand and structure. In the initial one the ligand was placed on the protein surface, next to the entrance to the active site (at ~16 Å from the heme’s iron atom), and allowed to diffuse freely into the heme distal site. The structure with the best (ligand) binding energy from the initial simulation was then selected for a second local refinement run, where the ligand center of mass was constrained to move within 8 Å from the heme’s iron. All simulations used 50 epochs of 16 MC PELE steps each with 128 computing cores. Interaction/binding energies (kcal/mol) were derived as Eab - (Ea + Eb), where Eab is the total energy of the complex, Eb the energy of the ligand and Ea the energy of receptor (everything but the ligand), all of them obtained at the OPLS2005 level of theory with a surface GB implicit solvent model.

Results and discussion

Regioselectivity in isophorone transformation by UPOs

In the present work, the ability of several UPOs to oxidize isophorone was analyzed, and different transformation patterns were observed (Fig. 1). The peroxygenases from the ascomycetes C. globosum (CglUPO) and H. insolens (rHinUPO), and the basidiomycete A. aegerita (AaeUPO) were found to transform the substrate, although with different regioselectivities (Table 1). In contrast, the enzymes from the basidiomycetes C. cinerea (rCciUPO) and M. rotula (MroUPO) were unable to convert the substrate, even at the highest (10 μM) enzyme doses tested.

101 5 Publications and Patents

IP

4HIP 7FIP 7HIP 4KIP AaeUPO rHinUPO CglUPO rCciUPO MroUPO Control 6 7 8 9 10 11 12 Retention time (min) Fig. 1. GC-MS analysis of 0.1 mM isophorone reactions (30 min) with 0.1 µM of MroUPO, rCciUPO, CglUPO, rHinUPO, AaeUPO and control without enzyme, showing the remaining substrate (IP, isophorone) and the hydroxylated (4HIP and 7HIP) and oxo (4KIP and 7FIP) derivatives.

CglUPO and rHinUPO selectively oxidized isophorone in position 4 (Scheme 1A) yielding mono-hydroxylated 4HIP and the keto derivative 4KIP, which were identified by their retention times compared with authentic standards (Fig. S1) and their mass spectra (Fig. S2). Besides the 4HIP and 4KIP molecular ions at m/z 154 and 152, respectively, their mass spectra showed a shift from α-isophorone fragments at m/z 54 and 82 to m/z 70 and 98 in 4HIP and to m/z 68 and 96 in 4KIP, which corresponds to the insertion of a hydroxyl or keto group. Isophorone conversion by rHinUPO and CglUPO (84 and 95%, respectively) was already efficient at low enzyme dose (0.1 μM) and both enzymes showed the tendency to over-oxidize 4HIP to form 4KIP. Over-oxidation was more pronounced in the case of rHinUPO as shown by the ratio of 4HIP to 4KIP of 1:1 (compared to 4:1 in the case of CglUPO) (Table 1).

AaeUPO was found to be less regioselective in oxidizing isophorone, since – in addition to 4HIP – other mono-hydroxylated and keto-derivatives were formed (Scheme 1B). These side products were identified as 7HIP and 7-formylisophorone (7FIP) due to their mass spectra (Fig. S3) that matched with those previously reported,34 with the singular mass fragment at m/z 125 in 7HIP, different from that at m/z 112 in 4HIP.

102 Publications and Patents 5

Table 1 Comparison of three UPOs in isophorone (0.1 mM) conversion (% of substrate) and relative abundance of the products 4HIP, 7HIP, 4KIP and 7FIP (% of the total products) after 30 min reaction, using different enzyme and peroxide doses. Conversion 4HIP 7HIP 4KIP 7FIP

AaeUPOa,d 72 46 25 tr 29 AaeUPOb,d 96 46 19 tr 35 CglUPOa,d 95 79 - 20 1 CglUPOc,e 100 - - 100 - rHinUPOa,d 84 53 - 45 2 rHinUPOc,e 100 3 - 97 - a-c Reactions using 0.1, 0.25 and 0.5 µM UPO, respectively. d,e Reactions using 2.5 and 5 mM H2O2, respectively; tr denotes traces

Reactions with low AaeUPO dose (0.1 μM) showed that hydroxylation took place in similar proportion at C4 (46%) and C7 (54%), but 7HIP was rapidly over- oxidized to form the corresponding aldehyde, and 7FIP and 4HIP were the major final products of the reaction (Table 1). Interestingly, 4HIP was barely further oxidized to 4KIP.

The regioselectivity observed in the hydroxylation of isophorone by some UPOs is similar to that reported for certain P450s. Among them, P450cam-RhFRed variants have been reported to yield 4HIP, 7HIP and isophorone oxide (2,3-epoxy- 3,5,5-trimethyl-1-cyclohexanone) as major products, with 4HIP as the only product from one of the variants.11 4HIP was also the main product of reactions with CYP102A1 and CYP101A1, although minor amounts of the epoxide, 7HIP and further oxidation products were observed.34 P450s, however, unlike rHinUPO or CglUPO, seem to be unable to oxidize 4HIP into 4KIP and, therefore, two enzymes (a P450 and an alcohol dehydrogenase) are necessary to obtain 4KIP from isophorone.11

103 5 Publications and Patents

O O O A 1 2 6 CglUPO/ CglUPO/ 3 5 HinUPO HinUPO 4 9 7 8 H2O2 H2O H2O2 H2O OH O IP 4HIP 4KIP

O O O B AaeUPO + HO H O H O 2 2 2 OH IP 4HIP 7HIP

H2O2 H2O2 AaeUPO AaeUPO H2O H2O

O O

O

O 4KIP 7FIP

Scheme 1. isophorone (IP) hydroxylation catalyzed by CglUPO and rHinUPO (A) and AaeUPO (B), showing the hydroxylated 4HIP (4-hydroxyisophorone) and 7HIP (7- hydroxyisophorone) and the oxo 4KIP (4-ketoisophorone) and 7FIP (7-formylisophorone) derivatives.

In view of the higher selectivity to form the products of interest (4HIP and 4KIP), CglUPO and rHinUPO were selected to perform reactions with higher (x100) substrate load (Fig. 2). These experiments revealed a faster substrate conversion by rHinUPO that completely transformed isophorone within 6 h, while CglUPO needed 12 h for 87% conversion. As expected, a higher proportion of 4KIP was observed in the rHinUPO reactions. A higher enzyme dose would be needed to complete conversion into 4KIP, as it was already found when lower substrate concentrations were tested (Table 1).

104 Publications and Patents 5

10 IP A 4HIP 4KIP

8 (mM) (mM) 6

4 Concentration 2

0 0 2 4 6 8 10 12 Time (h) B 10

8

6

4 Concentration (mM) Concentration 2

0 0 2 4 6 8 10 12 Time (h) Fig. 2. Time course of 10 mM isophorone (IP) reaction with 5 µM CglUPO (A) and -1 rHinUPO (B) and H2O2 (added with a syringe pump to give 5 mM h concentration), showing substrate and products (4HIP and 4KIP) concentrations.

Kinetics of isophorone hydroxylation by UPOs

Despite the difficulties to determine initial enzymatic reaction rates by GC-MS, kinetic curves for isophorone hydroxylation by the three UPOs could be obtained (Fig. S4) and reaction constants (kcat, Km and kcat/Km) were estimated (Table 2). There were differences in enzyme affinities, since the Km values were four- and two-fold higher for AaeUPO than for CglUPO and rHinUPO, revealing higher isophorone affinity of the two latter enzymes. Moreover, rHinUPO displayed a ten- fold higher turnover number (kcat) compared to CglUPO, which resulted in five-fold higher catalytic efficiency, while the efficiencies of AaeUPO and CglUPO were similar.

105 5 Publications and Patents

The catalytic efficiency of these UPOs hydroxylating isophorone is in the range of previously reported for the hydroxylation of cyclohexane by other UPOs.41 On the other hand, the Km of CglUPO and rHinUPO for isophorone hydroxylation is similar to values (380-440 μM) reported for CYP102A1 variants when decoy molecules were used, and the turnover numbers of these variants (2.5-5.5 s-1) were similar as well to that reported herein for CglUPO (4.4 s-1).34

Table 2 Kinetic parameters of AaeUPO, CglUPO and rHinUPO for the hydroxylation of isophoronea -1 -1 -1 kcat (s ) Km (µM) kcat/Km (mM ·s ) AaeUPO 20.1 ± 1.1 1 380 ± 200 14.6 ± 2.2 CglUPO 4.4 ± 0.4 309 ± 55 14.2 ± 2.8 rHinUPO 42.0 ± 9.8 633 ± 201 66.4 ± 26.1 a Data represent mean values of three replicates with standard deviations

Higher total turnover numbers (TTNs) were attained with rHinUPO (2 660) than with CglUPO (1 820) under the same reaction conditions, although these values could be up to 5 600 and 3 600 when lower enzyme doses were used (Table 3). The TTNs are in the range of those reported for two-step (1 567) or one-step (3 421) isophorone oxidation by the combination of P450-WAL and Cm-ADH10.11

Table 3 Product concentration and catalytic performance – given by total turnover number (TTN) and turnover frequency (TOF) – of CglUPO and rHinUPO after 9-h reaction using a higher isophorone concentration (10 mM) 4HIP (mM) 4KIP (mM) TTN TOF (min-1)

CglUPOb 6.7 1.2 1 820 3.4

CglUPOa 4.8 1.2 3 600 6.7 rHinUPOb 2.3 5.5 2 660 4.9 rHinUPOa 2.4 4.4 5 600 10.4

a,b Reactions using 2 and 5 µM enzyme, respectively

106 Publications and Patents 5

UPO enantioselectivity on isophorone and racemic 4HIP

Enantioselectivity in the synthesis of 4HIP by the three UPOs hydroxylating isophorone was determined by HPLC (Fig. S5). The results of the chiral analysis showed that only the reaction with AaeUPO can be considered as stereoselective, with an enantiomeric excess (ee) of 88% S-4HIP (Table 4). The three UPOs preferentially formed the S-enantiomer, in contrast to P450s that rather formed the R-enantiomer. 11,42

Table 4 Results from chiral HPLC analysis of the 10-mM isophorone reaction (60 min) with AaeUPO, CglUPO and rHinUPO (5 µM) showing the yields of the R and S enantiomers of 4HIP, and the resulting enantiomeric excess (ee), together with the amount of 4KIP formed and the total conversion yield (% of substrate) under the given conditionsa,b Conversion 4HIP 4KIP R-4HIP S-4HIP ee (%) (mM) (mM) (%) (%) (%) AaeUPO [a] 51 2.5 0.1 6 94 88 CglUPO [a] 24 3.3 0.3 30 70 40 rHinUPO [b] 36 2.0 0.4 48 52 4

a,b Reactions using 20 and 5 mM H2O2, respectively

The above values were estimated under reaction conditions where the product concentration was similar for all three enzymes (2.0-3.3 mM) and 4HIP over- oxidation was minimal, since it was observed that the ee of the hydroxylation product changed when 4KIP was formed. This was due to higher velocity in the conversion of the S-enantiomer compared to the R-enantiomer as observed in rHinUPO reactions with racemic 4HIP as substrate (Fig. S6). That way, a kinetic resolution of the racemate, with ee of 99-100% and 60-75% recovery, can be achieved with rHinUPO and CglUPO. In contrast, AaeUPO just slowly converted the racemic mixture of 4HIP, as it was also observed in the reaction of the enzyme with isophorone, where only traces of 4KIP were formed (Fig. 1), and no enantiomeric enrichment was produced.

107 5 Publications and Patents

Computational analyses: Molecular modeling

Isophorone diffusion was done in two simulation steps using PELE software.40 For the initial simulation, after preparing all systems placing isophorone on the surface next to the entrance channel, the substrate was allowed to move freely within 16 Å of the heme iron. In all systems, the substrate moved easily into the active site. Structures in direct contact with the heme produced the best binding- energy poses, although they presented different substrate orientations. Therefore, after selecting the best binding-energy structure, we run a local refinement, where the ligand was forced to move within ~8 Å of the heme iron. Fig. 3 shows an example of these two runs for CglUPO, where we display the binding-energy profile with respect to the C4-oxo distance for the initial (black) and the refinement (red) simulations.

)

mol Binding (kcal/ energy Binding CglUPO initial simulation CglUPO refinement simulation

IP-C to heme-oxo distance (Å) 4

Fig. 3. Example of the initial (black) and refinement (red) simulations on isophorone (IP) diffusion (C4-oxo distance being monitored with respect to binding/interaction energy) on CglUPO, using two-step adaptive PELE.35

108 Publications and Patents 5

Fig. 4A shows the binding energy along the C4-oxo distance for isophorone in the refinement runs for the five UPO structures (identified with different colors). Among them, three UPOs show distances lower than 3 Å: CglUPO (red), AaeUPO (green) and rHinUPO (cyan) (see Fig. S7 for the individual PELE plots). Such positioning will largely facilitate the hydrogen-atom abstraction by compound I, in agreement with the experimental results (Table 1). Interestingly, different isophorone reactive poses were detected for the different UPOs, as shown in Fig. 4B and C for CglUPO and AaeUPO, respectively. A closer examination of these structures showed shorter distances to the heme oxo for the pro-S (1.9 Å in both) than for the pro-R (3.5 and 2.6 Å, respectively, not shown) hydrogen atoms, which explained the preferential formation of the 4HIP S-enantiomer, as shown in Table 4. The resulting S-4HIP was also diffused with PELE, and strong differences in the C4-oxo distances and energies were obtained, as illustrated in Fig. 4D for two of the systems, with CglUPO closely approaching the heme oxo (Fig. 4E) while no catalytic distances were attained by AaeUPO (Fig. 4F). The above results agree with isophorone oxidation to 4KIP by CglUPO, while 4HIP is the main product from AaeUPO (Table 1), as well as with the deracemization results of chiral 4HIP (Fig. S6). On the other hand, although additional PELE calculations showed a similar C4-oxo distance for the R-enantiomer (data not shown), the slightly worse (5 kcal/mol) binding energy of the different pose adopted (with respect to S-4HIP) is in agreement with the S preference experimentally observed (Fig. S6).

Finally, the dual hydroxylation at the isophorone C4 and C7 positions by AaeUPO, compared with the selective oxidation at C4 by CglUPO and rHinUPO (Table 1) was also analyzed in the PELE simulations. The isophorone C7 position is not at a catalytically relevant distance in CglUPO (4.4 Å, Figs. 4B and S8A), while the C7-oxo and C4-oxo distances for AaeUPO (3.3 and 2.9 Å, respectively, Figs. 4C and S8B) are within reaction limits (the oxo to hydroxyl-H distance, in yellow, is also shown in Figs. 4B and C). The above results explain the lack of C7 hydroxylation by CglUPO and the similar percentages of C7-derivatives (7HIP+7FIP) and C4-derivatives (4HIP+4KIP) by AaeUPO (Table 1).

109 5 Publications and Patents

A D

CglUPO M roUPO Binding energy (kcal/mol)Bindingenergy CglUPO AaeUPO AaeUPO rHinUPO rCciUPO

IP-C4 to heme-oxo distance (Å) HIP-C4 to heme-oxo distance (Å)

Fig. 4. Isophorone (IP, left) and (S)-4-hydroxyisophorone (S-HIP, right) diffusion refinement on five UPOs, with adaptive PELE[35] monitoring the distance between the substrate and the oxo atom (red sphere) of the H2O2-activated heme with respect to the binding energy (Eb). A) C4 distance vs energy plot for IP diffusion in CglUPO (red), MroUPO (magenta), AaeUPO (green), rHinUPO (cyan) and rCciUPO (blue) (see Fig. S7 for individual PELE plots of the five UPO systems). B and C) IP at the two lowest Eb positions during the CglUPO and AaeUPO simulations (A), respectively. D) H4 distance vs energy plot for S-HIP diffusion on CglUPO (red) and AaeUPO (green). E and F) S-HIP at the two lowest Eb positions during the CglUPO and AaeUPO simulations (D), respectively.

110 Publications and Patents 5

Conclusions

We report a new enzymatic route for isophorone hydroxylation to form 4HIP and 4KIP, which are interesting products for the flavour-and-fragrance and pharmaceutical industries. The direct enzymatic oxidation of isophorone to 4KIP (with only one enzyme) is reported here for the first time for two fungal peroxygenases (CglUPO and rHinUPO). The above represents an advantage over the route with P450s, since the latter needs two enzymes (a P450 and an alcohol dehydrogenase) to obtain 4KIP from isophorone. However, process optimization of isophorone conversion by UPO is needed to attain the high-scale transformations reported for whole-cell P450 systems.43

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the EnzOx2 (H2020-BBI-PPP-2015-2-1-720297) EU- project, the AGL2014-53730-R (BIORENZYMERY) and CTQ2016-79138-R projects of the Spanish MINECO (co-financed by FEDER) and the CSIC (201740E071) project. Novozymes (Bagsvaerd, Denmark) is acknowledged for providing samples of rCciUPO and rHinUPO.

References

1. S. Krill, K. Huthmacher and S. Perrin, Patent (USA), 2002, 6469215 B1. 2. M. Eggersdorfer, D. Laudert, U. Létinois, T. McClymont, J. Medlock, T. Netscher and W. Bonrath, Angew. Chem. Int. Ed., 2012, 51, 12960-12990. 3. O. Isler, H. Lindlar, M. Montavon, R. Rüegg, G. Saucy and P. Zeller, HCA, 1956, 39, 2041-2053. 4. J. J. Becker, U. P. Hochstrasser and W. Skorianetz, Patent (USA), 1972, 3944620. 5. W. Zhong, L. Mao, Q. Xu, Z. Fu, G. Zou, Y. Li, D. Yin, H. Luo and S. R. Kirk, Appl. Cat. A: General, 2014, 486, 193-200. 6. M. Hennig, K. Püntener and M. Scalone, Tetrahedron: Asymmetry, 2000, 11, 1849-1858.

111 5 Publications and Patents

7. M. Ishihara, T. Tsuneya, H. Shiota, M. Shiga and K. Nakatsu, J. Org. Chem., 1986, 51, 491-495. 8. I. Kiran, O. Ozsen, T. Celik, S. Ilhan, B. Y. Gürsu and F. Demirci, Natural Product Communications, 2013, 8, 59-61. 9. Y. Mikami, Y. Fukunaga, M. Arita, Y. Obi and T. Kisaki, Agric. Biol. Chem., 1981, 45, 791-793. 10. I. Kaluzna, T. Schmitges, H. Straatman, D. van Tegelen, M. Müller, M. Schürmann and D. Mink, Org. Process Res. Dev., 2016, 20, 814-819. 11. M. Tavanti, F. Parmeggiani, J. R. G. Castellanos, A. Mattevi and N. J. Turner, ChemCatChem, 2017, 9, 3338-3348. 12. Y. Wang, D. Lan, R. Durrani and F. Hollmann, Curr. Opin. Chem. Biol., 2017, 37, 1-9. 13. R. Ullrich, J. Nuske, K. Scheibner, J. Spantzel and M. Hofrichter, Appl. Environ. Microbiol., 2004, 70, 4575-4581. 14. D. H. Anh, R. Ullrich, D. Benndorf, A. Svatos, A. Muck and M. Hofrichter, Appl. Environ. Microbiol., 2007, 73, 5477-5485. 15. G. Gröbe, M. Ullrich, M. Pecyna, D. Kapturska, S. Friedrich, M. Hofrichter and K. Scheibner, AMB Express, 2011, 1, 31-42. 16. J. Kiebist, K. U. Schmidtke, J. Zimmermann, H. Kellner, N. Jehmlich, R. Ullrich, D. Zänder, M. Hofrichter and K. Scheibner, ChemBioChem, 2017, 18, 563-569. 17. M. Hofrichter, H. Kellner, M. J. Pecyna and R. Ullrich, Adv. Exp. Med. Biol., 2015, 851, 341-368. 18. J. E. Stajich, S. K. Wilke, D. Ahren, C. H. Au, B. W. Birren, M. Borodovsky, C. Burns, B. Canbäck, L. A. Casselton, C. K. Cheng, J. X. Deng, F. S. Dietrich, D. C. Fargo, M. L. Farman, A. C. Gathman, J. Goldberg, R. Guigo, P. J. Hoegger, J. B. Hooker, A. Huggins, T. Y. James, T. Kamada, S. Kilaru, C. Kodira, U. Kües, D. Kupfert, H. S. Kwan, A. Lomsadze, W. X. Li, W. W. Lilly, L. J. Ma, A. J. Mackey, G. Manning, F. Martin, H. Muraguchi, D. O. Natvig, H. Palmerini, M. A. Ramesh, C. J. Rehmeyer, B. A. Roe, N. Shenoy, M. Stanke, V. Ter Hovhannisyan, A. Tunlid, R. Velagapudi, T. J. Vision, Q. D. Zeng, M. E. Zolan and P. J. Pukkila, Proc. Natl. Acad. Sci. USA, 2010, 107, 11889-11894. 19. C. Aranda, R. Ullrich, J. Kiebist, K. Scheibner, J. C. del Río, M. Hofrichter, A. T. Martínez and A. Gutiérrez, Catal. Sci. Technol., 2018, 8, 2394-2401. 20. M. Hofrichter and R. Ullrich, Curr. Opin. Chem. Biol., 2014, 19, 116-125.

112 Publications and Patents 5

21. A. Olmedo, C. Aranda, J. C. del Río, J. Kiebist, K. Scheibner, A. T. Martínez and A. Gutiérrez, Angew. Chem. Int. Ed., 2016, 55, 12248-12251. 22. E. D. Babot, J. C. del Río, L. Kalum, A. T. Martínez and A. Gutiérrez, Biotechnol. Bioeng., 2013, 110, 2332. 22. A. Gutiérrez, E. D. Babot, R. Ullrich, M. Hofrichter, A. T. Martínez and J. C. del Río, Arch. Biochem. Biophys., 2011, 514, 33-43. 23. S. Peter, M. Kinne, X. Wang, R. Ulrich, G. Kayser, J. T. Groves and M. Hofrichter, FEBS J., 2011, 278, 3667-3675. 24. C. Aranda, A. Olmedo, J. Kiebist, K. Scheibner, J. C. del Río, A. T. Martínez and A. Gutiérrez, ChemCatChem, 2018, 10, 3964-3968. 25. A. Olmedo, J. C. del Río, J. Kiebist, K. Scheibner, A. T. Martínez and A. Gutiérrez, Chem. -Eur. J., 2017, 23, 16985-16989. 26. E. D. Babot, J. C. del Río, M. Cañellas, F. Sancho, F. Lucas, V. Guallar, L. Kalum, H. Lund, G. Gröbe, K. Scheibner, R. Ullrich, M. Hofrichter, A. T. Martínez and A. Gutiérrez, Appl. Environ. Microbiol., 2015, 81, 4130-4142. 27. R. Ullrich, M. Poraj-Kobielska, S. Scholze, C. Halbout, M. Sandvoss, M. J. Pecyna, K. Scheibner and M. Hofrichter, J. Inorg. Biochem., 2018, 183, 84- 93. 28. F. Lucas, E. D. Babot, J. C. del Río, L. Kalum, R. Ullrich, M. Hofrichter, V. Guallar, A. T. Martínez and A. Gutiérrez, Catal. Sci. Technol., 2016, 6, 288- 295. 29. E. D. Babot, J. C. del Río, L. Kalum, A. T. Martínez and A. Gutiérrez, ChemCatChem, 2015, 7, 283-290. 30. R. Ullrich and M. Hofrichter, FEBS Lett., 2005, 579, 6247-6250. 31. H. Lund, L. Kalum, M. Hofrichter and S. Peter, Patent (USA), 2018, US 9908860 B2. 32. C. R. Otey, Methods Mol. Biol., 2003, 230, 137-139. 33. S. Dezvarei, J. H. Z. Lee and S. G. Bell, Enzyme Microb. Technol., 2018, 111, 29-37. 34. M. Biasini, S. Bienert, A. Waterhouse, K. Arnold, G. Studer, T. Schmidt, F. Kiefer, T. G. Cassarino, M. Bertoni, L. Bordoli and T. Schwede, Nucleic Acids Res., 2014, 42, W252-W258. 35. N. Guex, M. C. Peitsch and T. Schwede, Electrophoresis, 2009, 30, S162- S173. 36. G. M. Sastry, M. Adzhigirey, T. Day, R. Annabhimoju and W. Sherman, J. Comput. Aided Mol. Des., 2013, 27, 221-234.

113 5 Publications and Patents

37. V. Guallar and F. H. Wallrapp, Biophys. Chem., 2010, 149, 1-11. 38. A. Altun, V. Guallar, R. A. Friesner, S. Shaik and W. Thiel, J. Am. Chem. Soc., 2006, 128, 3924-3925. 39. D. Lecina, J. F. Gilabert and V. Guallar, Sci. Rep., 2017, 7, 8466. 40. S. Peter, M. Kinne, R. Ullrich, G. Kayser and M. Hofrichter, Enzyme Microb. Technol., 2013, 52, 370-376. 41. I. Kaluzna, T. Schmitges, H. Straatman, D. van Tegelen, M. Müller, M. Schürmann and D. Mink, Org. Process Res. Dev., 2016, 20, 814-819. 42. I. Kaluzna, T. Schmitges, H. Straatman, D. van Tegelen, M. Müller, M. Schürmann and D. Mink, Org. Process Res. Dev., 2016, 20, 814-819.

114 Publications and Patents 5

Supporting Information

Selective Synthesis of 4-Hydroxyisophorone and 4- Ketoisophorone by Fungal Peroxygenases

Carmen Aranda,a Martí Municoy,b Víctor Guallar,b,c Jan Kiebist,d Katrin Scheibner,d René Ullrich,e José C. del Río,a Martin Hofrichter,e Angel T. Martínez,*f and Ana Gutiérrez,*a

a Instituto de Recursos Naturales y Agrobiología de Sevilla, CSIC, Reina Mercedes 10, E-41012 Seville, Spain. E-mail: mail: [email protected] b Barcelona Supercomputing Center, Jordi Girona 31, E-08034, Barcelona, Spain. c ICREA, Passeig Lluís Companys 23, E-08010, Barcelona, Spain. d TU Dresden, Department of Bio- and Environmental Sciences, Markt 23, D- 02763 Zittau, Germany e JenaBios GmbH, Löbstedter Str. 80, D-07749 Jena, Germany f Centro de Investigaciones Biológicas, CSIC, Ramiro de Maeztu 9, E-28040 Madrid, Spain. E-mail: [email protected] * Corresponding author

115 5 Publications and Patents

4KIP A IP C 4KIP

4HIP

6 7 8 9 10 7 8 9 10 11

4KIP B D 4KIP

4HIP

6 7 8 9 10 7 8 9 10 11 Retention time (min) Retention time (min)

Figure S1. Comparison of GC-MS retention times of the products from isophorone (IP) reaction with CglUPO (A) and rHinUPO (C), compared with the corresponding 4KIP and 4HIP (from 4KIP chemical reduction) authentic standards (B and D).

116 Publications and Patents 5

m/z 82 A O

100% 82 m/z 54

[M+] 54 138 67 95 57 77 91 110121 140 50 75 100 125 150 m/z

m/z 98 B O m/z 70

100% 98

OH

70 112

+ 55 [M ] 7783 109 139 154 50 75 100 125 150 m/z

m/z 96 C O m/z 68 100% 68 96 O

[M+] 152 109 69 137 55 81 91 110124 50 75 100 125 150 m/z

Figure S2. Mass spectra of isophorone (A) and the products from the enzymatic reaction with CglUPO, 4HIP (B) and 4KIP (C).

117 5 Publications and Patents

A m/z 82 O m/z 54 100% 82

[M+] 54 138 67 95 57 77 91 110121 140 50 75 100 125 150 m/z

m/z 98 B O m/z 70 98 100% HO

+ 125 [M ] 55 6770 154 57 50 77 111121 139 50 75 100 125 150 m/z

m/z 96 C O m/z 68 100% 96 68 O

[M+] 109 152 53 79 69 81 103 124 137 50 75 100 125 150 m/z

Figure S3. Mass spectra of isophorone (A) and the products from enzymatic reaction 7HIP (B) and 7FIP (C).

118 Publications and Patents 5

A 4

3

)

-1

2

Turnover (s Turnover 1

0 0 1000 2000 3000 IP (µM) B 20

15

)

-1

10

Turnover (s Turnover 5

0 0 200 400 600 800 1000 1200 1400 1600 1800 IP (µM) C 20 18 16

) 14

-1 12 10 8

Turnover (s Turnover 6 4 2 0 0 2000 4000 6000 IP (µM)

Figure S4. Kinetic curves of enzymatic hydroxylation of isophorone (IP) by CglUPO (A), rHinUPO (B) and AaeUPO (C) from GC-MS estimation of 4HIP/4KIP formation (initial rates), adjusted as described in Experimental.

119 5 Publications and Patents

IP 100% 3x104 O A IP O

4HIP OH

Intensity OH S-4HIP Response R-4HIP 4KIP 4KIP

7 8 9 10 11 10 15 20 25 27

IP IP

B 100% 3x104 Intensity

Response 4HIP 4KIP 4KIP R-4HIP S-4HIP

7 8 9 10 11 10 15 20 25 27

C 100% IP 3x104

4HIP 7HIP IP

Response S-4HIP Intensity 7-FIP 7FIP 4KIP 4KIP R-4HIP

7 8 9 10 11 10 15 20 25 27

D 100% 3x104

4HIP

Response Intensity R-4HIP S-4HIP

7 8 9 10 11 10 15 20 25 27 Retention time (min) Retention time (min)

Figure S5. GC-MS (left) and chiral HPLC (right) analyses of isophorone (IP) hydroxylation by CglUPO (A), rHinUPO (B) and AaeUPO (C) and the chemical reduction of 4KIP (D), showing the R-4HIP, S-4HIP, 4KIP and 7FIP products.

120 Publications and Patents 5

100 A

80 (%) S 60

R substrate 40

20 Remaining

0 0 5 15 30 60 Time (min)

100 B

(%) 80

60 substrate

40

20 Remaining

0 0 5 15 30 60 Time (min)

100 C

80 (%)

60 substrate 40

20 Remaining

0 0 5 15 30 60 Time (min)

Figure S6. R-4HIP and S-4HIP enantiomers during reaction of 4- hydroxyisophorone (4-HIP) racemate with rHinUPO (A), CglUPO (B) and AaeUPO (C), in percentage (%) of the initial chiral substrate.

121 5 Publications and Patents

A D

) mol

CglUPO

MroUPO rHinUPO Binding (kcal/energy Binding Binding (kcal/mol) energy Binding AaeUPO rHinUPO rCciUPO

IP-C4 to heme-oxo distance (Å) IP-C4 to heme-oxo distance (Å)

B E

) )

mol mol

Binding (kcal/energy Binding Binding (kcal/energy Binding

AaeUPO CglUPO

IP-C4 to heme-oxo distance (Å) IP-C4 to heme-oxo distance (Å)

C F

) )

mol mol

Binding (kcal/energy Binding Binding (kcal/energy Binding rCciUPO MroUPO

IP-C4 to heme-oxo distance (Å) IP-C4 to heme-oxo distance (Å)

Figure S7. Individual PELE plots for isophorone (IP) diffusion in AaeUPO (B), rCciUPO (C), rHinUPO (D), CgllUPO (E) and MroUPO (F) showing the C4-oxo distance vs the binding energy, compared with the overlapping plots shown in Figure 4A (A).

122 Publications and Patents 5

A B Binding Binding energy (kcal/mol)

C4-oxo CglUPO C4-oxo AaeUPO C7-oxo CglUPO C7-oxo AaeUPO

IP-(C4/C7) to heme-oxo distance (Å) IP-(C4/C7) to heme-oxo distance (Å)

Figure S8. Comparison of C4- and C7-oxo distances vs binding energy during isophorone (IP) diffusion on CglUPO (A) and AaeUPO (B) using adaptive [41] PELE. For the same binding energy, the C7 distances are always shown by black dots, while the C4 distances for CglUPO and AaeUPO are shown by red and green dots, respectively.

123

Publications and Patents 5

H2O2 H2O

O O O

OR OR

5.1. Publication III: Aranda C., Olmedo A., Kiebist J., Scheibner K., del Río J.C., Martínez A.T. and Gutiérrez A. (2018). Selective epoxidation of fatty acids and fatty acid methyl esters by fungal peroxygenases. ChemCatChem 10: 3964-3968.

125

Publications and Patents 5

Selective Epoxidation of Fatty Acids and Fatty Acid Methyl Esters by Fungal Peroxygenases

Carmen Aranda1‡, Andrés Olmedo1‡, Jan Kiebist2, Katrin Scheibner2, José C. del Río1, Angel T. Martínez3, and Ana Gutiérrez*1

1 Instituto de Recursos Naturales y Agrobiología de Sevilla, CSIC, Reina Mercedes 10, 41012 Seville (Spain)

2 JenaBios GmbH Orlaweg 2, 00743 Jena (Germany)

3 Centro de Investigaciones Biológicas, CSIC Ramiro de Maeztu 9, 28040 Madrid (Spain)

* Corresponding author

‡ These authors contributed equally to this work

Keywords: epoxidation • fatty acids • fatty acid methyl esters fungal peroxygenases • oxidoreductases

Abstract:

Recently discovered fungal unspecific peroxygenases from Marasmius rotula and Chaetomium globosum catalyze the epoxidation of unsaturated fatty acids (FA) and FA methyl esters (FAME), unlike the well-known peroxygenases from Agrocybe aegerita and Coprinopsis cinerea. Reactions of a series of unsaturated FA and FAME with cis-configuration revealed high (up to 100%) substrate conversion and selectivity towards epoxidation, although some significant differences were observed between enzymes and substrates with the best results being obtained with the C. globosum enzyme. This and the M. rotula peroxygenase appear as promising biocatalysts for the environmentally-friendly production of reactive FA epoxides given their self-sufficient monooxygenase activity and the high conversion rate and epoxidation selectivity.

127 5 Publications and Patents

Oils and fats are among the most important renewable feedstock of the chemical industry, whose possibilities are still far from being fully exploited.[1] By simple industrial operations, fatty acids (FA) are available from vegetable oils in such purity that they may be used for further chemical transformations. Their conversion to FA methyl esters (FAME) is a well-known application of fats and oils, largely investigated for biodiesel production. Moreover, unsaturated FA and FAME can be further epoxidized, and used in industrial syntheses of chemicals and intermediates.

The industrial-scale epoxidation of unsaturated FA compounds is generally [2] carried out by the Prileshajev reaction via percarboxylic acids (Scheme 1a2). However, this method, which often includes strong mineral acids as catalysts for the "in situ" generation of peracids (Scheme 1a1), suffers from several drawbacks such as the relatively low selectivity for epoxides due to oxirane ring opening in the acidic medium, the corrosive nature of acids, and the unstable character of peracids.[3] Many studies have been aimed at searching an alternative, such as the chemo-enzymatic synthesis with lipases catalyzing the carboxylic acid reaction with hydrogen peroxide.[4;5] However, the latter reaction maintains most drawbacks of peracid-based epoxidation. Therefore, direct enzymatic processes emerge as an alternative solution for more selective and environmentally friendly epoxidation of unsaturated lipids. Several enzymes are known to catalyze epoxidation directly, such as cytochrome P450 monooxygenases (P450), diiron-center oxygenases, and plant peroxygenases.[5;6] However, they present some drawbacks, such as their intracellular nature, and the requirement for costly co-substrates in the two former cases.

Here, we show a promising enzymatic technology to epoxidize unsaturated FA (Scheme 1b) under mild and environmentally-friendly conditions, as potential alternative to the above chemical and enzymatic epoxidations. This includes the use of two recently discovered unspecific peroxygenases (UPO, EC 1.11.2.1), from the fungi Marasmius rotula (MroUPO)[7] and Chaetomium globosum (CglUPO).[8] These and related fungal peroxygenases represent a new class of enzymes that eludes some of the limitations of other monooxygenases since they are secreted proteins, therefore far more stable, and only require H2O2 for activation.

128 Publications and Patents 5

Conventional chemical route O O + (a ) H 1 R1 C OH + H2O2 R1 C O OH + H2O

O O O + (a2) R2 C C R3 + R1 C O OH R2 C C R3 R1 C OH H H H H This work: new enzymatic route O UPO (b) R2 C C R3 + H2O2 R2 C C R3 + H2O H H H H

Scheme 1. Chemical and enzymatic routes for the epoxidation of fatty acids.

Peroxygenases are structurally related to the P450s, as they also contain the heme prosthetic group coordinated by a cysteine ligand, but they do not depend on the reductive activation of molecular oxygen and catalyze the transfer of an oxygen atom from peroxide to substrates.[9] Initially, these UPO were shown to catalyze oxygenation reactions on aromatic compounds,[10] and their action on aliphatic compounds was demonstrated later.[11] After the first UPO discovered in Agrocybe aegerita (AaeUPO),[12] similar enzymes were found in other basidiomycetes, such as M. rotula, and there are indications for their widespread occurrence in the fungal kingdom.[13] Over one-hundred peroxygenase-type genes have been identified in the analysis of 24 basidiomycete genomes,[14] including Coprinopsis cinerea.[15] One UPO from the latter fungus is produced as a recombinant protein (rCciUPO) by Novozymes (Bagsvaerd, Denmark) (Figure S1A).[16] Interestingly, the recently described MroUPO presents differences with the best studied fungal peroxygenases, such as the ability of oxidizing bulkier substrates,[9;17] the terminal hydroxylation of n-alkanes[18] and the chain-shortening of carboxylic acids.[19] On the other hand, CglUPO is the fourth wild-type described UPO, and the first isolated from an ascomycete.[8]

129 5 Publications and Patents

The reactions of purified MroUPO and CglUPO (Figure S1B,C) with a series of cis-monounsaturated FA (from C14:1 to C22:1), showed that both enzymes are capable of oxygenating these substrates, CglUPO being more active since it achieved maximal substrate conversion with lower enzyme doses (Table 1). Interestingly, both peroxygenases generated the epoxidized derivatives as main products (Figures 1A-B), unlike the other well-known AaeUPO (not shown) and rCciUPO that were not able to epoxidize the double bond and instead, produced the hydroxyderivatives at the subterminal positions (Figure 1C). Therefore, whereas AaeUPO and rCciUPO shows similar regioselectivity towards saturated and unsaturated FA, MroUPO and CglUPO behave differently, hydroxylating at the terminal and/or subterminal positions the saturated FA and oxygenating the double bonds of unsaturated ones. The hydroxylation of FA at the subterminal positions by AaeUPO and rCciUPO was previously described.[11;16] However, the epoxidation of a fatty acid by a fungal peroxygenase is revealed here for the first time. Moreover, whereas CglUPO was highly selective with all unsaturated FA, over 90% epoxidation except with erucic (C22:1) acid, epoxidation by MroUPO depended on FA chain length, showing the highest value with myristoleic (C14:1) acid.

Besides epoxides, minor amounts of other products were found with CglUPO and especially with MroUPO, such as oxygenated derivatives of epoxidized FA (ED) and hydroxylated derivatives of FA (HD) mainly at terminal or subterminal positions of the carbon chain, and at the allylic positions (Table 1, Figure 1B). The higher efficiency and selectivity of CglUPO than MroUPO for epoxidation of most FA is shown in reactions with oleic (C18:1) acid (Figures S5-S6). Curiously, erucic acid, an abundant fatty acid in rapeseed and mustard oils, was transformed and epoxidized at a larger extent by MroUPO. The high selectivity of these UPOs epoxidizing oleic and palmitoleic (C16:1) acids (up to 100%) differs from that of P450 (BM3) where hydroxylation (> 97% and 65%, respectively) predominated over epoxidation (< 3% and 35%).[6]

In addition to monounsaturated FA, some polyunsaturated FA (linoleic and α- and γ-linolenic acids) abundant in vegetable oils were tested as substrates. Although both UPOs transformed almost completely linoleic acid at the highest enzyme doses (Table 2), CglUPO was more selective producing the diepoxide (both syn and anti- enantiomers) in very high yield (92% of total products) (Figure 2A). MroUPO, besides the epoxides also generated hydroxylated derivatives of the monoepoxides

130 Publications and Patents 5

and oxygenated derivatives at allylic positions (mainly at -7) (Figure 2B). When lower doses of enzymes were used, a predominance of monoepoxides over diepoxides was observed (Table 2).

Table 1. Monounsaturated FA (R=H) and FAME (R=CH3) reactions with CglUPO and MroUPO doses yielding maximal conversion into epoxides, together with other oxygenated (hydroxyl, keto and carboxyl) derivatives at different positions (arrows) O O O O UPO O O ( ) ( ) + ( ) + ( ) m m m m ( ) ( )n OR ( )n OR ( )n OR n OR H2O2 H2O E ED HD

Products (%) H O Tot Substrate Enzyme dose 2 2 E ED HD (mM) (µM)

R=H 14:1 cis-Δ9 CglUPO 60 nM 5 99 - 1 98 (n=5, m=1) MroUPO 200 nM 5 86 8 6 99 16:1 cis-Δ9 CglUPO 60 nM 5 91 4 5 99 (n=5, m=3) MroUPO 200 nM 2.5 65 17 18 99 18:1 cis-Δ9 CglUPO 60 nM 2.5 91 2 7 99 (n=5, m=5) MroUPO 200 nM 5 38 61 1 96 20:1 cis-Δ11 CglUPO 60 nM 5 95 4 1 94 (n=7, m=5) MroUPO 400 nMa 2.5 62 7 31 99 22:1 cis-Δ13 CglUPO 250 nM 2.5 50 12 39 77* (n=9, m=5) MroUPO 400 nMc 5 67 10 23 91

R=CH3 14:1 cis-Δ9 CglUPO 60 nM 2.5 100 - - 94 MroUPO 200 nM 2.5 68 13 19 100 18:1 cis-Δ9 CglUPO 1 µMb 5 98 - 2 75* MroUPO 200 nMc 2.5 73 7 21 93 Substrates (100 µM), enzyme doses, and estimated total products (µM), relative abundance (% of total products) of epoxide (E), epoxide derivatives (ED), hydroxylated derivatives (HD) are shown. Reactions were performed in 20% acetone, 30 min at 30ºC except for reactions a,b and c that were performed in 40% acetone at 40ºC, for 30 min, 60 min and 120 min, respectively. *Higher enzyme concentration (up to 500 nM) did not improve conversion. See Figure S2 for GC-MS of authentic standards and Figures S3A,S4 for mass spectra of E and ED.

131 5 Publications and Patents

A E

C18:1 cis-9

18 20 22 24 26 28 30 Retention time (min)

B E ED

(-7)-OH

(-1)-OH

keto

-

1)

-

 ( C18:1 cis-9 -OH COOH

18 20 22 24 26 28 30 Retention time (min)

HD C (-2)-OH

(-1)-OH C18:1 cis-9

18 20 22 24 26 28 30 Retention time (min)

Figure 1. GC-MS of reactions of oleic acid (underlined) at 30 min with 60 nM CglUPO (A), 200 nM MroUPO (B), and 100 nM rCciUPO (C), showing the epoxide (E), epoxide derivatives (ED) and the hydroxylated derivatives (HD) of oleic acid.

132 Publications and Patents 5

Table 2. Oxidation of linoleic acid and its methyl ester by CglUPO and MroUPO

O O O O O O O + + OR OR OR O 12-E ED di-E UPO OR O O O O O H2O2 H2O + + OR OR OR 9-E ED HD

Products (%)

H2O2 12-E 9-E di-E ED HD Total Enzyme dose (mM) (µM) R=H CglUPO 30 nM 2.5 39 57 2 - 2 76 125 nM 5 - - 92 8 - 98 MroUPO 100 nM 2.5 37 25 10 25 3 75 400 nM 2.5 - - 39 59 2 95 R=CH3 CglUPO 250 nM 5 13 59 28 - - 80 1 µM 5 - - 100 - - 100 MroUPO 200 nM 2.5 39 40 4 8 9 95 1 µM 5 - - 49 51* - 98 Substrate (100 µM), enzyme doses, H2O2 conc., amount of estimated total products (µM) and relative abundance (% of total products) of 12-epoxide (12-E), 9-epoxide (9-E), di- epoxide (di-E), epoxide derivatives (ED) and hydroxylated derivatives (HD) are shown. Arrows indicate the main chain positions oxidized by the enzymes in epoxidized derivatives (ED) and hydroxylated derivatives (HD). Reactions conditions: 20% acetone at 30ºC, 30 min (R=H), 40% acetone at 40ºC, 60 min (CglUPO) and 120 min (MroUPO) (R=CH3).* Mono- and di-epoxide derivative. See Figure S7 for GC-MS of authentic standards and Figures S3B, S8 for mass spectra of E and ED.

Linolenic acids (α- and γ-) were also transformed by MroUPO and CglUPO (Figure S9). Both, monoepoxides (located in two different double bonds) and diepoxides were generated from α-linolenic acid under the conditions tested. However, only one epoxide was observed in the γ-linolenic acid reactions.

To compare the efficiency of MroUPO and CglUPO oxidizing unsaturated FA, apparent kinetic constants were determined for oleic acid oxidation (Table 3) in spite of the difficulties for GC- MS estimation of initial reaction rates. Regarding the turnover rate, CglUPO presented three-fold higher kcat values than MroUPO. In addition, the Km value was four-fold higher for MroUPO, which represented less

133 5 Publications and Patents

affinity by this enzyme compared to CglUPO. As a result, one order of magnitude higher catalytic efficiency (kcat/Km) was observed with CglUPO. In agreement with these results, CglUPO presented total turnover numbers (TTN) up to 8000 in oleic acid reactions when the substrate concentration was increased to 1 mM while this number was half for MroUPO with 50 nm enzyme being used in both cases.

A di-E-syn

di-E-anti

ED C18:2 cis,cis-Δ9,Δ12

18 20 22 24 26 28 30 Retention time (min)

B di-E-syn

di-E-anti ED

ED

(-7)-OH-9-E (-7)-OH-12-E

C18:2 cis,cis-Δ9,Δ12

18 20 22 24 26 28 30 Retention time (min) Figure 2. GC-MS of reactions of linoleic acid (underlined) at 30 min with 125 nM CglUPO (A) and 400 nM MroUPO (B), showing the diepoxides (di-E), epoxide derivatives (ED) and the hydroxylated derivatives of linoleic acid (see mass spectrum of the diepoxide in Figure S3B).

134 Publications and Patents 5

This is a promising value accompanied by a significantly higher product amount (of about 0.5 mM) considering the limited solubility and other difficulties for fatty acid epoxidation. Likewise, the turnover frequency (TOF) was double for CglUPO (2.2 s-1) than for MroUPO (1.1 s-1). Solubility limitations prevented calculation of accurate kinetic constants for linoleic acid oxidation, since saturation could not be estimated especially for MroUPO (Figure S10), but higher activity than found for oleic acid was observed at high linoleic acid concentration.

FAME, usually obtained from vegetable oils by transesterification with methanol, were also tested as substrates of MroUPO and CglUPO. Namely, the methyl esters of two monounsaturated (myristoleic and oleic acids) and one diunsaturated FA (linoleic acid) were selected. Both peroxygenases were able to transform and epoxidize the monounsaturated FAME (Figures 3A,C). CglUPO showed similar selectivity towards the esters than with the free FA, but differences in the case of oleic acid were observed with MroUPO (Table 1). Regarding the methyl ester of linoleic acid, CglUPO showed a strict selectivity towards epoxidation (generating the diepoxide) while MroUPO was less selective towards diepoxide formation (Table 2, Figures 3B,D). In contrast, P450 BM3, which hydroxylate/epoxidize free fatty acids, was reported as unable to hydroxylate FAME [20]. This seems related to the fact that the free carboxyl group is required to fix the substrate at the entrance of P450 active site [6]. Finally, it is interesting that different patterns of oxygenation were observed with the cis isomers of the substrates (compared to the trans isomers).

Table 3. Estimated kinetic parameters for oleic acid oxidation by CglUPO and MroUPO. Data represent mean values of three replicates with standard errors

-1 -1 -1 kcat (s ) Km (µM) kcat/ Km (M ·s )

CglUPO 8.1 ± 0.9 10.7 ± 4.0 7.6 ± 3.0 x 105 MroUPO 2.6 ± 0.2 38.9 ± 6.1 6.7 ± 1.1 x 104

135 5 Publications and Patents

While MroUPO converted predominantly myristelaidic, palmitelaidic and elaidic acids (or their methyl esters) into the hydroxyderivatives at the allylic positions, CglUPO generated mainly the epoxides, with the only exception of elaidic acid (data not shown).

E di-E-anti di-E-syn A B

9 CH3-C18:1cis-Δ 9 12 (-7)-OH CH3-C18:2 cis,cis-Δ ,Δ

18 20 22 24 16 18 20 22 24 Retention time (min) Retention time (min)

E di-E-syn C D ED di-E-anti

(-7)-OH ED 9 12 (-10)-OH (-10)-OH (-7)-OH CH3-C18:2 cis,cis-Δ ,Δ 9 CH3-C18:1cis-Δ

18 20 22 24 16 18 20 22 24 Retention time (min) Retention time (min)

Figure 3. GC-MS of reactions of methyl oleate (left, underlined) with 1 µM CglUPO at 60 min (A) and 200 nM MroUPO at 120 min and (C), showing the epoxide (E), epoxide derivatives (ED); and the hydroxylated derivatives, and methyl linoleate (right, underlined) with 1 µM CglUPO at 60 min (B) and 1 µM MroUPO at 120 min (D), showing the diepoxides (di-E), epoxide derivatives (ED) and the hydroxylated derivatives of methyl linoleate.

136 Publications and Patents 5

The selective epoxidation of FA and FAME, a reaction of great interest for the chemical industry,[21] must be added to the repertoire of UPOs, as dream biocatalysts for oxyfunctionalization chemistry.[9;22;23] The structural determinants driving to selective epoxidations in MroUPO and CglUPO (compared to AaeUPO and CciUPO) are difficult to be identified with the information available on these new heme-thiolate enzymes (note that only one UPO crystal structure has been published to date). However, in related P450, epoxidation vs hydroxylation rates have been related to the balance between the iron hydroperoxo and oxenoid forms after the oxidative activation of the enzyme, with an active site threonine being involve in the transition as the proton donor.[24;25] Interestingly, a threonine residue is present at the active sites of both AaeUPO and CciUPO, and absent from those of MroUPO and CglUPO, as shown by Aranda et al.[26] but its relevance in the FA hydroxylation/epoxidation balance is still to be experimentally investigated. This and other structural-functional studies with UPOs will help to understand the reaction mechanisms of these versatile enzymes, and to obtain ad-hoc variants for biotechnological application.

Acknowledgements

This work was funded by the BIORENZYMERY (AGL2014-53730-R) project of the Spanish MINECO (co-financed by FEDER) and by the CSIC 201740E071 project. H. Lund (Novozymes) is acknowledged for rCciUPO.

References

[1] U. Biermann, U. Bornscheuer, M. A. R. Meier, J. O. Metzger, H. J. Schäfer, Angew. Chem. Int. Ed. 2011, 50, 3854-3871. [2] N. Prileschajew, Ber. Dtsch. Chem. Ges. 1909, 42, 4811-4815. [3] S. M. Danov, O. A. Kazantsev, A. L. Esipovich, A. S. Belousov, A. E. Rogozhin, E. A. Kanakov, Catal. Sci. Technol. 2017, 7, 3659-3675. [4] F. Björkling, H. Frykman, S. E. Godtfredsen, O. Kirk, Tetrahedron 1992, 48, 4587-4592. [5] C. Tiran, J. Lecomte, E. Dubreucq, P. Villeneuve, OCL 2008, 15, 179-183. [6] R. T. Ruettinger, A. J. Fulco, J. Biol. Chem. 1981, 256, 5728-5734.

137 5 Publications and Patents

[7] G. Gröbe, M. Ullrich, M. Pecyna, D. Kapturska, S. Friedrich, M. Hofrichter, K. Scheibner, AMB Express 2011, 1, 31-42. [8] J. Kiebist, K. U. Schmidtke, J. Zimmermann, H. Kellner, N. Jehmlich, R. Ullrich, D. Zänder, M. Hofrichter, K. Scheibner, ChemBioChem 2017, 18, 563-569. [9] M. Hofrichter, H. Kellner, M. J. Pecyna, R. Ullrich, Adv. Exp. Med. Biol. 2015, 851, 341-368. [10] M. Hofrichter, R. Ullrich, M. J. Pecyna, C. Liers, T. Lundell, Appl. Microbiol. Biotechnol. 2010, 87, 871-897. [11] A. Gutiérrez, E. D. Babot, R. Ullrich, M. Hofrichter, A. T. Martínez, J. C. del Río, Arch. Biochem. Biophys. 2011, 514, 33-43. [12] R. Ullrich, J. Nuske, K. Scheibner, J. Spantzel, M. Hofrichter, Appl. Environ. Microbiol. 2004, 70, 4575-4581. [13] M. J. Pecyna, R. Ullrich, B. Bittner, A. Clemens, K. Scheibner, R. Schubert, M. Hofrichter, Appl. Microbiol. Biotechnol. 2009, 84, 885-897. [14] D. Floudas, M. Binder, R. Riley, K. Barry, R. A. Blanchette, B. Henrissat, A. T. Martínez, R. Otillar, J. W. Spatafora, J. S. Yadav et al., Science 2012, 336, 1715-1719. [15] J. E. Stajich, S. K. Wilke, D. Ahren, C. H. Au, B. W. Birren, M. Borodovsky, C. Burns, B. Canbäck, L. A. Casselton, C. K. Cheng et al., Proc. Natl. Acad. Sci. USA 2010, 107, 11889-11894. [16] E. D. Babot, J. C. del Río, L. Kalum, A. T. Martínez, A. Gutiérrez, Biotechnol. Bioeng. 2013, 110, 2332. [17] E. D. Babot, J. C. del Río, M. Cañellas, F. Sancho, F. Lucas, V. Guallar, L. Kalum, H. Lund, G. Gröbe, K. Scheibner, R. Ullrich, M. Hofrichter, A. T. Martínez, A. Gutiérrez, Appl. Environ. Microbiol. 2015, 81, 4130-4142. [18] A. Olmedo, C. Aranda, J. C. del Río, J. Kiebist, K. Scheibner, A. T. Martínez, A. Gutiérrez, Angew. Chem. Int. Ed. 2016, 55, 12248-12251. [19] A. Olmedo, J. C. del Río, J. Kiebist, K. Scheibner, A. T. Martínez, A. Gutiérrez, Chem. Eur. J. 2017, 23, 16985-16989. [20] Y. Miura, A. J. Fulco, Biochim. Biophys. Acta-Lipids Lipid Metab. 1975, 388, 305-317. [21] A. Corma, S. Iborra, A. Velty, Chem. Rev. 2007, 107, 2411-2502. [22] Y. Wang, D. Lan, R. Durrani, F. Hollmann, Curr. Opin. Chem. Biol. 2017, 37, 1-9.

138 Publications and Patents 5

[23] S. Bormann, A. G. Baraibar, Y. Ni, D. Holtmann, F. Hollmann, Catal. Sci. Technol. 2015, 5, 2038-2052. [24] A. D. N. Vaz, D. F. McGinnity, M. J. Coon, Proc. Natl. Acad. Sci. USA 1998, 95, 3555-3560. [25] E. G. Hrycay, S. M. Bandiera, Adv. Exp. Med. Biol. 2015, 851, 1-61. [26] C. Aranda, R. Ullrich, J. Kiebist, K. Scheibner, J. C. del Río, M. Hofrichter, A. T. Martínez, A. Gutiérrez, Catal. Sci. Technol. 2018, 8, 2394-2401.

139 5 Publications and Patents

Supporting Information

© Copyright Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim, 2018

Selective Epoxidation of Fatty Acids and Fatty Acid Methyl Esters by Fungal Peroxygenases

Carmen Aranda+, Andrés Olmedo+, Jan Kiebist, Katrin Scheibner, José C. del Río, Angel T. Martínez, and Ana Gutiérrez*

140 Publications and Patents 5

Table of contents

1. Supplemental materials and methods 1.1 Enzymes 1.2 Model substrates 1.3 Enzymatic reactions 1.4 Enzyme kinetics 1.4 GC-MS analyses 2. Supplemental figures 3. Supplemental references 4. Complete reference cited in the main text

1. Supplemental materials and methods

1.1. Enzymes

MroUPO is a wild enzyme isolated from cultures of M. rotula DSM 25031, a fungus deposited at the German Collection of Microorganisms and Cell Cultures Braunschweig, Germany). It was purified by fast protein liquid chromatography (FPLC) to apparent homogeneity, confirmed by sodium dodecylsulfate- polyacrylamide gel electrophoresis under denaturing conditions, and showed a molecular mass of 32 kDa and isoelectric point of pH 5.0-5.3. The UV-visible spectrum of the enzyme showed a characteristic maximum at 418 nm (Soret band of heme-thiolate proteins).[1]

CglUPO (36 kDa) is a wild enzyme isolated from cultures of C. globosum DSM 62110, from the German Collection of Microorganisms and Cell cultures. It was purified by ammonium sulfate precipitation and successive FPLC on Q-Sepharose FF, Superdex75, and Mono Q columns using an ÄKTA FPLC system (GE Healthcare).[2]

rCciUPO (used with comparative purpose) was provided by Novozymes A/S (Bagsvaerd, Denmark). This recombinant enzyme corresponds to the protein model 7249 from the sequenced C. cinerea genome available at the JGI (http://genome.jgi.doe.gov/Copci1), expressed in Aspergillus oryzae (patent WO/2008/119780), and purified using a combination of S-Sepharose and SP- Sepharose ion-exchange chromatography. The recombinant peroxygenase

141 5 Publications and Patents

preparation is an electrophoretically homogeneous glycoprotein with a molecular mass around 44 kDa, a typical UV-vis spectrum, and the ability to oxygenate different aromatic compounds with a specific activity of approximately 100 U·mg- 1 (measured as described below). SDS PAGE of different UPO preparations are shown in Fig. S1.

One UPO activity unit is defined as the amount of enzyme oxidizing 1 µmol of -1 -1 veratryl alcohol to veratraldehyde (ε310 9300 M ·cm ) in 1 min at 24ºC, pH 7, after addition of 2.5 mM H2O2. Enzyme concentration was estimated from the characteristic spectrum of peroxidase complex with carbon monoxide. [3]

1.2. Model substrates

A series of unsaturated lipids (cis isomers) from Sigma-Aldrich was used including: i) fatty acids such as myristoleic (cis-9-tetradecenoic), palmitoleic (cis- 9-hexadecenoic), oleic (cis-9-octadecenoic), linoleic (cis,cis-9,12- octadecadienoic), α- (cis, cis, cis-9,12,15), and γ-(cis, cis, cis-6, 9,12) linolenic (octadecatrienoic) acids, gondoic (cis-11-eicosenoic) and erucic (cis-13- docosenoic) acids; and vi) fatty acid esters such as methyl myristoleate, methyl oleate and methyl linoleate. The trans isomers such as myristelaidic (trans-9- tetradecenoic), palmitelaidic (trans-9-hexadecenoic) and elaidic (trans-9- octadecenoic) acids were also obtained from Sigma-Aldrich. The following standards of epoxides were also used, rac cis-9,10-epoxyoctadecanoic acid and (±)9(10)-EpOME (9,10-cis-epoxide of linoleic acid) from Santa Cruz Biotechnology; (±)12(13)-EpOME (12,13-cis-epoxide of linoleic acid) from Cayman; and 9,10-12,13-diepoxyoctadecanoic acid from Larodan.

1.3. Enzymatic reactions

Reactions of the model compounds at 0.1 mM concentration were performed with: i) MroUPO (50-400 nM) in 50 mM sodium phosphate buffer (pH 5.5) at 30ºC or 40ºC and 30-120 min reaction time, in the presence of 0.5-5 mM H2O2 and ii) CglUPO (30-1000 nM) in 50 mM sodium phosphate buffer (pH 7) at 30ºC or 40ºC and 30-60 min reaction time, in the presence of 0.5-5 mM H2O2. In all cases the H2O2 was added in pulses. Prior to use, the substrates were dissolved in acetone and added to the buffer to give a final acetone concentration of 20% (v/v) although concentrations of 40% were also tested with some compounds. In control experiments, substrates were treated under the same conditions (including 2.5-5

142 Publications and Patents 5

mM H2O2) but without enzyme. Products were recovered by liquid-liquid extraction with methyl tert-butyl ether and dried under N2. N,O-bis (trimethylsilyl)trifluoroacetamide (Supelco) was used to prepare trimethylsilyl (TMS) derivatives that were analyzed by GC-MS.

1.4. Enzyme kinetics

To study the kinetics of fatty acids (oleic and linoleic acid) oxidation, reactions in 1 mL vials with 50 nM of enzyme (MroUPO and CglUPO) were carried out. Substrate concentration was varied between 6.25 μM and 400 μM and 20% (v/v) of acetone was used as cosolvent. The reactions were initiated with 0.5 mM H2O2 and stopped after 1 min (CglUPO) or 5 min (MroUPO) with 100 μL of 100 mM sodium azide solution by vigorous shaking. Time course of the reactions were previously followed to ensure the estimation of kinetic parameters while the reaction rate was still in linear range. All reactions were carried out in triplicate. Products quantification was performed by GCMS (as described below) using external standard curves and response factors of authentic standards. Kinetic parameters (kcat, Km) were obtained by fitting the data to the Michaelis-Menten equation using SigmaPlot software (Systat Softwarwe Inc., San Jose, CA, USA). For TTN (total turnover number) determination, substrate concentration was increased to 1 mM and 5 mM using methyl-β-cyclodextrin (from Sigma-Aldrich) in a final reaction concentration of 5 mM or 20% (v/v) of acetone using 50 nM of both enzymes. The highest product concentration was attained using acetone in the case of CglUPO (1 mM substrate, 400 μM products) and methyl-β-cyclodextrin in the case of MroUPO (1mM substrate, 200 μM products).

1.5. GC-MS analyses

The analyses were performed with a Shimadzu GC-MS QP2010 Ultra, using a fused silica DB-5HT capillary column (30 m x 0.25 mm internal diameter, 0.1 μm film thickness) from J&W Scientific. The oven was heated from 120°C (1 min) to 300°C (15 min) at 5°C·min-1. The injection was performed at 300°C and the transfer line was kept at 300°C. Compounds were identified by mass fragmentography, and comparing their mass spectra with those of the Wiley and NIST libraries and standards. Quantification was obtained from total-ion peak area, using molar response factors of the same or similar compounds.

143 5 Publications and Patents

2. Supplemental figures

Figure S1. SDS PAGE of different UPO preparations: rCciUPO (A), MroUPO (B) and CglUPO (C). 10-12% Bis-Tris was used, and the proteins were visualized with a colloidal Blue staining (Invitrogen). Conditions (50 mM dithiothreitol) resulted in monomeric MroUPO. Low molecular weight standards (Thermo Scientific, Darmstadt, Germany) were included.

144 Publications and Patents 5

Figure S2. GC-MS chromatograms of cis-9,10-epoxystearic acid (E) from oleic acid (underlined) reaction with CglUPO (1A) and cis-9,10-epoxystearic acid standard (2A); and of cis-9,10-epoxystearic acid methyl ester (E) from oleic acid methyl ester (underlined) reaction with CglUPO (1B) and cis-9,10-epoxystearic acid methyl ester standard (2B).

145 5 Publications and Patents

Figure S3. Mass spectra and formula/fragmentation of oleic acid epoxide (A) and linoleic acid diepoxide (B) (as trimethylsilyl derivatives) from CglUPO and MroUPO reactions with oleic acid and linoleic acid, respectively.

146 Publications and Patents 5

Figure S4. Mass spectra and formula/fragmentation of compounds tentatively assigned as hydroxylated derivative at (-7) position (A) and at (-1) position (B) of oleic acid epoxide (as trimethylsilyl derivatives) from MroUPO reactions with oleic acid (Fig. 1B).

147 5 Publications and Patents

Figure S5. GC-MS of conversion of oleic acid (underlined) by the same dose (50 nM) of CglUPO (black) and MroUPO (red), within 5 min, and 0.5 mM H2O2, showing the epoxide (E), epoxide derivatives (ED) and the hydroxylated derivatives (HD) of oleic acid at allylic positions (see mass spectra of the epoxide in Figures S3).

148 Publications and Patents 5

Figure S6. Molar percentage of compounds from reactions of oleic acid with 50 nM CglUPO and 100 nM MroUPO, under different final H2O2 concentrations and reaction times, showing the remaining substrate, epoxide (E), hydroxylated derivatives of the epoxide at (-1) and (-7) positions and the hydroxylated derivatives (HD) of oleic acid at subterminal positions.

149 5 Publications and Patents

Figure S7. GC-MS chromatograms of 12,13-cis-epoxide of linoleic acid (12-E); 9,10-cis-epoxide of linoleic acid (9-E) and 9,10-12,13-diepoxyoctadecanoic acid (di-E-anti and di-E-syn) from linoleic acid (underlined) reaction with CglUPO (1A) and of authentic standards cis-12,13-epoxystearic acid (2A); cis-9,10-epoxystearic acid (3A); and 9,10-12,13-diepoxyoctadecanoic acid (4A) as well as their corresponding methyl esters (1B-4B).

150 Publications and Patents 5

Figure S8. Mass spectra and formula/fragmentation of compounds tentatively assigned as hydroxylated derivatives at (-7) position of linoleic acid 12-epoxide (A) and 9-epoxide (B) (as trimethylsilyl derivatives) from MroUPO reactions with linoleic acid (Fig. 2B).

151 5 Publications and Patents

Figure S9. GC-MS of reactions of α-linolenic acid (left) at 30 min with 125 nM CglUPO (A) and 100 nM MroUPO (C), and γ-linolenic acid (right) at 30 min with 60 nM CglUPO (B) and 60 nM MroUPO (D), showing the remaining substrates (underlined), monoepoxides (E), diepoxides (di-E) and hydroxylated derivatives.

152 Publications and Patents 5

Figure S10. Kinetic studies for the enzymatic oxidation of linoleic acid by CglUPO (A) and MroUPO (B).

153 5 Publications and Patents

3. Supplemental references

[1] G. Gröbe, M. Ullrich, M. Pecyna, D. Kapturska, S. Friedrich, M. Hofrichter, K. Scheibner, AMB Express 2011, 1, 31-42.

[2] J. Kiebist, K. U. Schmidtke, J. Zimmermann, H. Kellner, N. Jehmlich, R. Ullrich, D. Zänder, M. Hofrichter, K. Scheibner, ChemBioChem 2017, 18, 563- 569.

[3] C. R. Otey, Methods Mol. Biol. 2003, 230, 137-139.

4. Complete reference cited in the main text

[14] D. Floudas, M. Binder, R. Riley, K. Barry, R. A. Blanchette, B. Henrissat, A. T. Martínez, R. Otillar, J. W. Spatafora, J. S. Yadav, A. Aerts, I. Benoit, A. Boyd, A. Carlson, A. Copeland, P. M. Coutinho, R. P. de Vries, P. Ferreira, K. Findley, B. Foster, J. Gaskell, D. Glotzer, P. Górecki, J. Heitman, C. Hesse, C. Hori, K. Igarashi, J. A. Jurgens, N. Kallen, P. Kersten, A. Kohler, U. Kües, T. K. A. Kumar, A. Kuo, K. LaButti, L. F. Larrondo, E. Lindquist, A. Ling, V. Lombard, S. Lucas, T. Lundell, R. Martin, D. J. McLaughlin, I. Morgenstern, E. Morin, C. Murat, M. Nolan, R. A. Ohm, A. Patyshakuliyeva, A. Rokas, F. J. Ruiz-Dueñas, G. Sabat, A. Salamov, M. Samejima, J. Schmutz, J. C. Slot, F. St.John, J. Stenlid, H. Sun, S. Sun, K. Syed, A. Tsang, A. Wiebenga, D. Young, A. Pisabarro, D. C. Eastwood, F. Martin, D. Cullen, I. V. Grigoriev, D. S. Hibbett, Science 2012, 336, 1715-1719.

154 Publications and Patents 5

5.2. Patent I:

Aranda Oliden C., del Río Andrade J. C., Martínez Ferrer A. T., Gutiérrez Suárez A. (2018). Process for the selective synthesis of 4-hydroxyisophorone and 4- ketoisophorone by fungal peroxygenases. EP 18382872.

155

Publications and Patents 5

157 5 Publications and Patents

158 Publications and Patents 5

159 5 Publications and Patents

160 Publications and Patents 5

5.2. Patent II:

Fernández Fueyo E., Aranda Oliden C., Gutiérrez Suárez A., Martínez Ferrer A. T. (2018). Method of heterologous expression of active fungal unspecific peroxygenase in bacterial host cells for fatty-acid epoxidation and other oxygenation reactions. EP 18382514.

161

Publications and Patents 5

163 5 Publications and Patents

164 Publications and Patents 5

165 5 Publications and Patents

166

6 Conclusions

Conclusions 6

6. Conclusions

Three different oxyfunctionalization reactions catalyzed by several fungal unspecific peroxygenases (UPOs), such as those from Agrocybe aegerita (AaeUPO), Marasmius rotula (MroUPO), Chaetomium globosum (CglUPO) and the recombinant enzymes from Coprinopsis cinerea (rCciUPO) and Humicola insolens (rHinUPO), are described for the first time in this Thesis. The main conclusions reached during this Thesis were:

1. All UPOs tested (AaeUPO, MroUPO, rCciUPO and CglUPO) were able to selectively transform trans-stilbene, although the peroxygenases from the basidiomycetes produced the resveratrol analogue 4,4′-dihydroxy-trans- stilbene (DHS) while the peroxygenase from the ascomycete produced trans-stilbene epoxide. AaeUPO was the most efficient enzyme catalyzing the double hydroxylation of trans-stilbene, as it was shown by kinetic studies and the higher catalytic performance exposed. Kinetic parameters also revealed that the monohydroxylated derivative 4-hydroxy-trans- stilbene was the best two electrons oxidation substrate of UPOs reported to date.

2. True peroxygenative mechanism was demonstrated in trans-stilbene 18 oxygenation in studies using H2 O2 as co-substrate, where the mass spectra of reaction products suffered the corresponding shifts indicating 18O incorporation.

3. Structural differences, such as the presence of four phenylalanine residues at the entrance of the heme-access channel in AaeUPO, compared to MroUPO and rCciUPO that have only one, might explain the higher efficiency of the first enzyme facilitating the substrate positioning by aromatic interactions. In addition, the very different distal side of the heme pocket in CglUPO with a unique and potentially reactive tyrosine residue in the middle of the active site could be the responsible for the very different regioselectivity observed.

171 6 Conclusions

4. AaeUPO, CglUPO and rHinUPO catalyzed the hydroxylation of α- isophorone, unlike MroUPO and rCciUPO that were unable to transform the substrate. CglUPO and rHinUPO selectively hydroxylated α-isophorone in C4 forming the flavour and fragrances molecules 4-hydroxyisophorone and 4-ketoisophorone, with higher amounts of the latter product in rHinUPO reactions. AaeUPO was the only stereoselective enzyme in 4- hydroxyisophorone synthesis with high enantiomeric excess for the S- enantiomer, although it was the less selective enzyme, since it also produced 7-hydroxyisophorone and 7-formylisophorone in addition to the C4 oxidized derivatives.

5. CglUPO and rHinUPO had more affinity to α-isophorone than AaeUPO, as shown by the lower values of the Michaelis-Menten constants that, together with the higher turnover number observed for rHinUPO, made this the most efficient biocatalyst for α-isophorone hydroxylation.

6. Computational studies explained the different reactivities observed for the different UPOs in α-isophorone hydroxylation, due to the different distances between the C4 of the substrate and the oxo-heme Compound I. The closer distance to the pro-S hydrogen would also explain the preferential S enantiomer formation of UPOs. On the other hand, the longer distances of C4 of 4-hydroxyisophorone to the oxo-heme of AaeUPO could be the responsible of the lack of activity towards this position.

7. MroUPO and CglUPO catalyzed the epoxidation of mono- and poly- unsaturated fatty acids and methyl esters. CglUPO was the most efficient biocatalyst with one order of magnitude higher catalytic efficiency in oleic acid epoxidation, and was also the most selective in the formation of the epoxide compared to MroUPO, with the only exception of erucic acid.

In conclusion, the comparison of UPOs performance carried out during this Thesis is important for the selection of the best (more efficient and selective) biocatalyst for the processes optimization. The three different biotransformation reactions described herein represent a more environmentally friendly alternative to chemical synthesis.

172

7 Annex

Annex 7

7.1. Publication:

Olmedo A., Aranda C., del Río J.C., Kiebist J., Scheibner K., Martínez A.T. and Gutiérrez A. (2016). From alkanes to carboxylic acids: terminal oxygenation by a fungal peroxygenase. Angewandte Chemie International Edition 55:12248-12251.

177

Annex 7

From Alkanes to Carboxylic Acids: Terminal Oxygenation by a Fungal Peroxygenase

Andrés Olmedo1, Carmen Aranda1, José C. del Río1, Jan Kiebist2, Katrin Scheibner2, Angel T. Martínez3, and Ana Gutiérrez*1

1 Instituto de Recursos Naturales y Agrobiología de Sevilla, CSIC, Reina Mercedes 10, 41012 Seville (Spain)

2 JenaBios GmbH Orlaweg 2, 00743 Jena (Germany)

3 Centro de Investigaciones Biológicas, CSIC Ramiro de Maeztu 9, 28040 Madrid (Spain)

* Corresponding author

Keywords: alkanes • carboxylic acids • oxidoreductases • peroxygenase • terminal hydroxylation

Abstract:

A new heme–thiolate peroxidase catalyzes the hydroxylation of n-alkanes at the terminal position—a challenging reaction in organic chemistry—with H2O2 as the only cosubstrate. Besides the primary product, 1-dodecanol, the conversion of dodecane yielded dodecanoic, 12-hydroxydodecanoic, and 1,12-dodecanedioic acids, as identified by GC–MS. Dodecanal could be detected only in trace amounts, and 1,12-dodecanediol was not observed, thus suggesting that dodecanoic acid is the branch point between mono- and diterminal hydroxylation. Simultaneously, oxygenation was observed at other hydrocarbon chain positions (preferentially C2 and C11). Similar results were observed in reactions of tetradecane. The pattern of products formed, together with data on the incorporation of 18O from the cosubstrate 18 H2 O2, demonstrate that the enzyme acts as a peroxygenase that is able to catalyze a cascade of mono- and diterminal oxidation reactions of long-chain n-alkanes to give carboxylic acids.

179 7 Annex

The selective oxyfunctionalization of saturated hydrocarbons under mild conditions is a major challenge in modern chemistry. Among the thousands of reagents for organic synthesis, few have been developed that are capable of the selective oxidation of alkanes.[1] The alkane C-H bond is extremely inert and difficult to hydroxylate. Additionally, the similarity of methylene C-H bond strengths in a linear alkane and the lack of functional groups that can direct catalysis make selective hydroxylation of these compounds highly challenging. On the basis of their relative bond strengths, the terminal methyl C-H bonds are inherently more difficult to oxidize than the secondary or tertiary C-H bonds in the hydrocarbon chain. Members of the cytochrome P450 monooxygenase (P450) superfamily catalyze the selective oxyfunctionalization of many organic substrates under mild and environmentally friendly conditions,[2] and some of them are able to catalyze the terminal oxygenation of alkanes.[3,4] However, owing to their frequent requirement for costly cosubstrates and auxiliary enzymes, among other reasons, applications of these versatile biocatalysts mainly focus on the production of drug metabolites, pharmaceutical products, and some specialty chemicals.[2,5,6]

A new heme peroxidase type was discovered 12 years ago in the basidiomycete Agrocybe aegerita,[7] which efficiently transfers oxygen to various organic substrates.[8,9] This enzyme is able to catalyze reactions formerly assigned only to P450s.[10] It differs from classical by the presence of a cysteine residue as the fifth ligand of the heme iron atom,[11] and shares the heme–thiolate center with P450s and with the chloroperoxidase from the ascomycete Leptoxyphium fumago, which also has oxygenation activity.[8] However, unlike P450s, which are intracellular enzymes, whose activation often requires an auxiliary enzyme or protein domain and a source of reducing power, the A. aegerita enzyme is a secreted protein. It is therefore far more stable, and more importantly only requires H2O2 for activation.[8] In the latter sense, peroxygenase catalysis has similarities with the so- called “peroxide shunt” operating in P450s, and with a few P450s that show strictly peroxide dependent activity.[12] However, basidiomycete peroxygenases generally have better catalytic and stability properties than the above peroxide-activated P450s.

The A. aegerita peroxygenase was shown to catalyze interesting oxygenation reactions on aromatic compounds, and more recently its action on aliphatic compounds was demonstrated,[13–16] thus expanding its biotechnological interest. Therefore, the enzyme is known as an unspecific peroxygenase (UPO). After the

180 Annex 7

first peroxygenase from A. aegerita (AaeUPO),[7] similar enzymes have been found in other basidiomycetes, such as Coprinellus radians (CraUPO)[17] and Marasmius rotula (MroUPO),[18] and there are indications for their widespread occurrence in the fungal kingdom.[19,20] Moreover, an UPO from the sequenced genome of Coprinopsis cinerea (CciUPO) has been expressed in an industrial host and shown to catalyze interesting hydroxylation reactions.[15,21,22] UPOs could approach the catalytic versatility of P450s and suitably supplement them in the near future.[8] However, there are a number of reactions that had not yet been shown for UPOs, including terminal alkane hydroxylation.[8] Previous studies[13,14,22] showed the hydroxylation of n-alkanes by AaeUPO and CciUPOs, but the reaction is always subterminal (Figure 1).

Alk(C14) 2-OH 3-OH

2,12-diOH 3,12-diOH 2,13-diOH

10 12 14 16 18 20 22 Retention time (min)

Figure 1. GC-MS analysis of AaeUPO reaction with tetradecane showing the remaining substrate (Alk, alkane) and the subterminal mono/di-dihydroxylated (OH) derivatives. See Supplemental material and methods for details.

The recently described MroUPO presents differences with respect to the most extensively studied UPOs, such as higher activity towards aliphatic compounds, as well as the ability to oxidize bulkier substrates,[8] and only shares approximately 30% sequence identity. It was also known that MroUPO presents differences in the active site, such as a histidine residue (instead of a conserved arginine residue) as a

181 7 Annex

charge stabilizer for heterolytic cleavage of the H2O2 O-O bond (after transient proton transfer to a conserved glutamate residue), thus resulting in compound I (CI) [8,11] plus H2O, although their relevance in catalysis is still to be established. Stimulated by these differences, we investigated the oxidation of n-alkanes with this new UPO.

With this purpose, we tested two linear saturated long-chain alkanes, n- dodecane and n-tetradecane, as MroUPO substrates and identified the oxygenation products by GC–MS. With a substrate concentration of 0.3 mM (in 20% acetone), 68 and 45% conversion of dodecane and tetradecane, respectively, was observed at 120 min in reactions with MroUPO (0.5 µM). Under these conditions, the enzyme is completely stable. The products of the reaction with dodecane are shown in Figure 2A (see also Table S1 in the Supporting Information), including those only formed by terminal hydroxylation/s, such as 1-dodecanol, dodecanoic acid, - hydroxydodecanoic acid, and 1,12-dodecanedioic acid. All the intermediates from an alkane 1 to a dicarboxylic acid 10 via the terminal fatty alcohol 2 and -hydroxy fatty acid 5 (Figure 3, left) were identified in the MroUPO reactions, except the monoaldehyde 3 (traces) and carboxyaldehyde 9, apparently as a result of their rapid further oxidation. However, no terminal diol 6, -hydroxyaldehyde 7, or dialdehyde 8 (Figure 3, right) were observed. One explanation is that conversion of the diol (if formed) into the diacid is favored to such a degree that it proved impossible to observe the aldehydes. Indeed, the rapid conversion of the diol into the diacid was observed in the reaction of dodecanediol (Figure 2D) and tetradecanediol (not shown), and no dialdehyde was observed. However, the possibility that the diol 6 is not formed and the dicarboxylic acid is only produced via the monocarboxylic acid 4 seems more feasible, since in the reaction of dodecanol (Figure 2B) only dodecanoic acid and its derivatives were identified. Indeed, the pattern of products derived from dodecanol is similar to that for dodecanoic acid (Figure 2C).

Some of the terminal-oxygenation products showed additional oxygenation at subterminal (-1 and -2) positions, with the formation of hydroxy and keto fatty acids (Figure 2A; see also Table S1). Therefore, in contrast with the exclusively subterminal hydroxylation reactions of n-alkanes by other UPOs (Figure 1), MroUPO is able to catalyze their terminal hydroxylation (ca. 50% of products in Figure 2). Moreover, a few products only showing subterminal oxygenation were also identified as alkane hydroxy, keto, and hydroxy/keto derivatives.

182 Annex 7

Alk (C12) A

COOH

(-1)-keto-COOH OH

2 - keto -

11 (-1)-OH-COOH - ()-OH-COOH

2- OH keto - 1- OH 2 di-COOH

10 15 20 Retention time (min) B COOH

(-1)-keto-COOH

(-1)-OH-COOH Alc (C12) ()-OH-COOH di-COOH

10 15 20 Retention time (min) di-COOH C

(-1)-keto-COOH

COOH

-

OH

-

COOH

-

1) -

 OH

-

( )

Ac (C12)  (

10 15 20 Retention time (min) D di-COOH

()-OH-COOH

di- Alc (C12) Ald-COOH

10 15 20 Retention time (min)

Figure 2. GC–MS analysis of MroUPO reactions with dodecane (A), 1-dodecanol (B), dodecanoic acid (C), and 1,12-dodecanediol (D) showing the remaining substrate (Alk, alkane; Alc, alcohol; and Ac, acid) and the terminal (bold), terminal/subterminal (bold, italics), and subterminal hydroxylated (OH) keto and carboxylic (COOH) derivatives (see the Supporting Information for details).

183 7 Annex

1 ( ) n

MroUPO H2O2 b 2 ( ) ( ) 6 n OH HO n OH

O c O 3 7 ( ) ( ) n H HO n H

O O O 4 ( ) ( ) 8 n OH H n H a O

5 ( ) HO n OH

O O

9 ( ) H n OH

O O 10 ( ) HO n OH

Figure 3. Pathways for the terminal oxygenation of n-alkanes to dicarboxylic acids, including identified and hypothetical intermediates, and three possible branch points (a, b and c) between mono- and dioxygenated compounds.

When the alkane reactions were performed at higher concentrations of acetone (40–60%) to improve solubility, the proportion of the compounds formed varied (see Table S1), probably as a result of increased relative solubility of the substrates with respect to oxidized intermediates. Finally, it was noted that higher conversion (up to 100%) was observed at lower substrate concentrations (as shown for 0.1 mM tetradecane in Figure S1 in the Supporting Information).

184 Annex 7

The most characteristic property of UPO is its ability to transfer oxygen to substrate molecules, which in the present case includes a cascade of sequential mono- and diterminal reactions of n-alkanes to give dicarboxylic acids. We therefore investigated the origin of the oxygen atoms introduced into the alkanes and intermediate compounds. The results of 18O labeling reactions revealed that an 18 oxygen atom from H2 O2 (90% isotopic purity) is introduced into n-tetradecane to form 1-tetradecanol, whose diagnostic fragment (m/z 271; Figure 4A, top) appeared fully (90%) 18O-labeled (m/z 273; Figure 4A, bottom). Direct evidence for the 18 incorporation of an oxygen atom from H2 O2 in aldehyde formation could not be obtained, since the aldehyde was barely detected.

However, 18O was incorporated in the carboxyl group of myristic acid, whose characteristic fragments (at m/z 285 and 117; Figure 4B, top) became 18O-bilabeled 18 (m/z 289 and 121; Figure 4B, bottom). Likewise, H2 O2 oxygen atoms were incorporated into the fatty acid -hydroxylated derivative (see Figure S2A: 18O- trilabeled diagnostic fragments at m/z 379 and 363) and dicarboxylic acid (see Figure S2B: 18O-tetralabeled diagnostic fragment at m/z 395). In summary, the 18 18 reaction of tetradecane in the presence of H2 O2 showed O labeling of the different hydroxy and carboxyl groups (see the Supporting Information for details). Therefore, it can be concluded that all oxygen atoms incorporated during alkane 18 oxidation by MroUPO are supplied by H2O2 and not from O2. The O-labeling results agree with the peroxygenation mechanism depicted below,[8,9] whereby the 3+ resting enzyme (RS), containing Fe and a porphyrin (P), is activated by H2O2 to yield CI, a Fe4+=O porphyrin cation radical (P•+) complex [Eq. (1)].

3+ •+ 4+ P-Fe (RS) + H2O2 → P -Fe =O(CI) + H2O (1)

•+ 4+ 4+ • P -Fe =O(CI) + RH → P-Fe =O(CII) + R (2)

4+ • 3+ P-Fe =O(CII) + R → P-Fe (RS) + ROH (3)

185 7 Annex

CI abstracts one H atom from the substrate (RH) to yield a radical (R•) plus compound II (CII), a Fe4+=O reduced porphyrin complex [Eq. (2)]. Finally, CII completes the hydroxylation reaction (R-OH formation) and returns to RS [Eq. (3)]. The initial product of n-alkane oxidation by MroUPO will be a terminal fatty alcohol, which is reported herein for the first time for a peroxygenase reaction.[8,13] The product of fatty-alcohol oxidation by the peroxygenase will be a gem-diol from a second C1 hydroxylation, and will be either i) directly hydroxylated (even at the nascent stage) to yield a gem-triol intermediate, with irreversible dehydration to release the fatty acid, or ii) first dehydrated to the aldehyde and then hydroxylated to the fatty acid. Most 18O-labeling data indicate that the gem-diol/aldehyde is immediately hydroxylated (without hydroxyl exchange with the solvent); however, the existence of minor simple labeling of the carboxyl group in some MroUPO reactions (together with double labeling) suggests some hydroxyl exchange with the water at the aldehyde/gem-diol stage (only aldehyde traces found in the chromatograms), although the loss of 18O labeling is much lower than reported for the P450 cascade oxidation of hexadecanol.[3] Hydroxylation of the aldehyde form was the mechanism suggested for the AaeUPO oxidation of benzyl alcohol to benzoic acid, in which, in contrast with our observations during this study, a substantial amount of the aldehyde accumulated.[23] Finally, no diols or dialdehydes were detected in alkane oxidation by MroUPO, unlike in the oxidation of n- hexadecane with P450.[3] Moreover, the pattern of products identified in the alkane (and fatty-alcohol) reactions suggests that diterminal oxyfunctionalization by MroUPO initiates at the monocarboxylic acid (Figure 3, reaction a) and not at the 1-alcohol or aldehyde (Figure 3, reactions b and c, respectively). Herein, we have described the first reaction cascade leading to reactive carboxylic acids from chemically inert alkanes with a peroxygenase. Having demonstrated the feasibility of the enzymatic terminal oxyfunctionalization of alkanes with MroUPO, we expect further studies to improve the regioselectivity of the enzyme, whose structure has been recently solved (PDB entry 5FUJ on hold), as reported for an engineered P450 BM3 variant that shows approximately 50% selectivity in the hydroxylation of the terminal position of a medium-chain alkane.[24] This peroxygenase type has high industrial potential for the mild activation of alkanes, with the advantages of self- sufficient monooxygenase activity, thus enabling large-scale transformations, and the ability to hydroxylate the most unreactive terminal positions.

186 Annex 7

+ [M - CH3] 100% 271 16OTMS A m/z 103

75

97 83 103 57 125 [M+] 286

50 100 150 200 250 300 m/z

+ [M - CH3] 100% 273 18OTMS m/z 105

77

57 97 105

125 [M+] 288 50 100 150 200 250 300 m/z

117 100% 16O B 73 16 OTMS + m/z 117 [M - CH3] 285

129 132 55 145 95 + 159 185 201 241 257 [M ] 300 50 100 150 200 250 300 m/z

100% 73 18O 18OTMS m/z 121 77 121

+ [M - CH3] 289 55 133 136 119 149 + 95 163 189 205 241 261 285 [M ] 304 50 100 150 200 250 300 m/z Figure 4. Mass spectra of 1-tetradecanol (A) and myristic acid (B) from MroUPO reactions with n-tetradecane in 18O-labeling experiments (bottom) and controls (top). The formulae 16 for the unlabeled compounds found in the H2 O2 reactions (A and B, top) and the labeled 18 compounds found in the H2 O2 (90% isotopic purity) reactions (A and B, bottom) are shown as trimethylsilyl (TMS) derivatives.

187 7 Annex

Acknowledgements:

This study was funded by the INDOX (KBBE-2013-7-613549) EU project, and the BIORENZYMERY (AGL2014-53730-R) and NOESIS (BIO2014-56388-R) projects of the Spanish MINECO (cofinanced by FEDER). R. Ullrich and M. Hofrichter (TU Dresden) are acknowledged for providing AaeUPO used for comparison, and E. D. Babot (IRNAS) for help in experimental assays.

References

[1] M. Bordeaux, A. Galarneau, J. Drone, Angew. Chem. Int. Ed. 2012, 51, 10712–10723; Angew. Chem. 2012, 124, 10870 – 10881.

[2] P. R. Ortiz de Montellano, Chem. Rev. 2010, 110, 932 – 948.

[3] U. Scheller, T. Zimmer, D. Becher, F. Schauer,W. H. Schunck, J. Biol. Chem. 1998, 273, 32528 – 32534.

[4] J. B. Johnston, H. Ouellet, L. M. Podust, P. R. Ortiz de Montellano, Arch. Biochem. Biophys. 2011, 507, 86 – 94.

[5] V. B. Urlacher, M. Girhard, Trends Biotechnol. 2012, 30, 26 – 36.

[6] R. Fasan, ACS Catal. 2012, 2, 647 – 666.

[7] R. Ullrich, J. Nuske, K. Scheibner, J. Spantzel, M. Hofrichter, Appl. Environ. Microbiol. 2004, 70, 4575 – 4581.

[8] M. Hofrichter, H. Kellner, M. J. Pecyna, R. Ullrich, Adv. Exp. Med. Biol. 2015, 851, 341 – 368.

[9] M. Hofrichter, R. Ullrich, Curr. Opin. Chem. Biol. 2014, 19, 116 –125.

[10] R. Bernhardt, J. Biotechnol. 2006, 124, 128 – 145.

[11] K. Piontek, E. Strittmatter, R. Ullrich, G. Grobe, M. J. Pecyna, M. Kluge, K. Scheibner, M. Hofrichter, D. A. Plattner, J. Biol. Chem. 2013, 288, 34767 – 34776.

[12] O. Shoji, Y. Watanabe, J. Biol. Inorg. Chem. 2014, 19, 529 – 539.

188 Annex 7

[13] A. Gutiérrez, E. D. Babot, R. Ullrich, M. Hofrichter, A. T. Martínez, J. C. del Río, Arch. Biochem. Biophys. 2011, 514, 33–43.

[14] S. Peter, M. Kinne, X. Wang, R. Ulrich, G. Kayser, J. T. Groves, M. Hofrichter, FEBS J. 2011, 278, 3667 – 3675.

[15] E. D. Babot, J. C. del Río, M. Cañellas, F. Sancho, F. Lucas, V. Guallar, L. Kalum, H. Lund, G. Gröbe, K. Scheibner, R. Ullrich, M. Hofrichter, A. T. Martínez, A. Gutiérrez, Appl. Environ. Microbiol. 2015, 81, 4130 – 4142.

[16] F. Lucas, E. D. Babot, J. C. del Río, L. Kalum, R. Ullrich, M. Hofrichter, V. Guallar, A. T. Martínez, A. Gutiérrez, Catal. Sci. Technol. 2016, 6, 288 – 295.

[17] D. H. Anh, R. Ullrich, D. Benndorf, A. Svatos, A. Muck, M. Hofrichter, Appl. Environ. Microbiol. 2007, 73, 5477 – 5485.

[18] G. Gröbe, M. Ullrich, M. Pecyna, D. Kapturska, S. Friedrich, M. Hofrichter, K. Scheibner, AMB Express 2011, 1, 31 – 42.

[19] M. J. Pecyna, R. Ullrich, B. Bittner,A. Clemens, K. Scheibner, R. Schubert, M. Hofrichter, Appl. Microbiol. Biotechnol. 2009, 84, 885 – 897.

[20] D. Floudas et al., Science 2012, 336, 1715 – 1719.

[21] E. D. Babot, J. C. del Río, L. Kalum, A. T. Martínez, A. Gutiérrez, ChemCatChem 2015, 7, 283 – 290.

[22] E. D. Babot, J. C. del Río, L. Kalum, A. T. Martínez, A. Gutiérrez, Biotechnol. Bioeng. 2013, 110, 2323.

[23] M. Kinne, C. Zeisig, R. Ullrich, G. Kayser, K. E. Hammel, M. Hofrichter, Biochem. Biophys. Res. Commun. 2010, 397, 18 – 21.

[24] P. Meinhold, M.W. Peters, A. Hartwick, A. R. Hernandez, F. Arnold, Adv. Synth. Catal. 2006, 348, 763 – 772.

[23] M. Kinne, C. Zeisig, R. Ullrich, G. Kayser, K. E. Hammel, M. Hofrichter, Biochem. Biophys. Res. Commun. 2010, 397, 18 – 21.

[24] P. Meinhold, M.W. Peters, A. Hartwick, A. R. Hernandez, F. Arnold, Adv. Synth. Catal. 2006, 348, 763 – 772.

189 Annex 7

Supporting Information

From Alkanes to Carboxylic Acids: Terminal Oxygenation by a Fungal Peroxygenase

Andrés Olmedo, Carmen Aranda, José C. del Río, Jan Kiebist, Katrin Scheibner, Angel T. Martínez, and Ana Gutiérrez*

anie_201605430_sm_miscellaneous_information.pdf

191 7 Annex

Table of contents 1. Supplemental materials and methods 1.1 Enzymes 1.2 Model substrates 1.3 Enzymatic reactions 1.4 GC-MS analyses 2. Supplemental results 2.1 n-Alkanes conversion: Product identification and effect of solvent concentration 2.2 Terminal hydroxylation of n-alkanes to dicarboxylic acids: 18O- labeling study 2.2.1 Fatty alcohol formation 2.2.2 Fatty acid formation 2.2.3 Hydroxyfatty acid formation 2.2.4 Dicarboxylic acid formation 3. Supplemental acknowledgement 4. Supplemental references 5. Complete reference cited in the main text

1. Supplemental materials and methods 1.1 Enzymes The MroUPO enzyme is a wild-type peroxygenase isolated from cultures of M. rotula DSM 25031, a fungus deposited at the German Collection of Microorganisms and Cell Cultures (Braunschweig, Germany). MroUPO was purified by FPLC to apparent homogeneity, confirmed by sodium dodecylsulfate- polyacrylamide gel electrophoresis under denaturing conditions, and showed a molecular mass of 32 kDa and isoelectric point of pH 5.0-5.3. The UV-visible spectrum of the enzyme showed a characteristic maximum at 418 nm (Soret band of heme-thiolate proteins).[1] All media and columns used for enzyme isolation were purchased from GE Healthcare Life Sciences.

192 Annex 7

The AaeUPO included in the present study for comparative purposes (A. aegerita isoform II, 46 kDa) was provided by R. Ullrich and M. Hofrichter (Technical University of Dresden, Germany) after its isolation from cultures of A. aegerita grown in soybean-peptone medium, and subsequent purification using a combination of Q-Sepharose and SP-Sepharose and Mono-S ion-exchange chromatographic steps[2]. One UPO activity unit is defined as the amount of enzyme oxidizing 1 mol -1 -1 of veratryl alcohol to veratraldehyde (ε310 9300 M ·cm ) in 1 min at 24ºC, pH 7 (the optimum for peroxygenase activity) after addition of 2.5 mM H2O2.

1.2 Model substrates Two alkanes, namely n-dodecane and n-tetradecane (from Sigma-Aldrich), were studied as substrates of MroUPO (and AaeUPO). Additionally, reactions using dodecanol and tetradecanol, dodecanoic and tetradecanoic acids (lauric and myristic acids, respectively) and dodecanediol and tetradecanediol as substrates (all of them from Sigma-Aldrich) were also studied to get further insight into the reactions of alkanes.

1.3 Enzymatic reactions Reactions of the two model alkanes at 0.3 mM concentration (except in Figure S1 where 0.1 mM was used for complete conversion) with MroUPO (0.5-1 µM) were performed in 5-mL vials containing 50 mM sodium phosphate (pH 5.5) at 40ºC and 30-120 min reaction time, in the presence of 2.5 mM H2O2, except for Figure S1 (10 mM), Figure 2A and dodecane reaction in Table S1 (5 mM in the two latter cases), where higher H2O2 concentrations yielded higher conversion rates. Prior to use, the substrates were dissolved in acetone and added to the buffer to give a final acetone concentration of 20% (v/v), although concentrations of 40% and 60% were also tested. A 120 min incubation in the presence of 20% acetone did not affect the activity of MroUPO while 40% and 60% acetone caused an activity loss of about 25%. In control experiments, substrates were treated under the same conditions (including 0.5-5 mM H2O2) but without enzyme. Enzymatic reactions 18 18 with O-labeled hydrogen peroxide (H2 O2, 90% isotopic content) from Sigma- Aldrich (2% w:v solution) were performed under the same conditions described

193 7 Annex

above. Likewise, MroUPO (0.5 µM) reactions with 0.1 mM of dodecanol and tetradecanol, dodecanoic and tetradecanoic acids and dodecanediol and tetradecanediol, 20% (v/v) acetone, and 2.5 mM H2O2 (incubated for 30-60 min) were also performed. Reactions of tetradecane (0.3 mM) with AaeUPO were performed in 40% (v/v) acetone, at pH 7 and 2.5 mM H2O2 (incubated for 120 min). Products were recovered by liquid-liquid extraction with methyl tert-butyl ether and dried under N2. N,O-Bis(trimethylsilyl)trifluoroacetamide (Supelco) was used to prepare trimethylsilyl (TMS) derivatives that were analyzed by GC-MS.

1.4 GC-MS analyses The analyses were performed with a Shimadzu GC-MS QP2010 Ultra, using a fused-silica DB-5HT capillary column (30 m x 0.25 mm internal diameter, 0.1 μm film thickness) from J&W Scientific. The oven was heated from 50°C (1.5 min) to 90°C (2 min) at 30°C·min-1, and then from 90°C to 250°C (15 min) at 8°C·min-1. The injection was performed at 250°C and the transfer line was kept at 300°C. Compounds were identified by mass fragmentography, and comparing their mass spectra with those of the Wiley and NIST libraries and standards. Quantification was obtained from total-ion peak area, using response factors of the same or similar compounds. Data from replicates were averaged and, in all cases (substrate conversion and relative abundance of reaction products), the standard deviations were below 3.5% of the mean values.

2. Supplemental results

2.1 n-Alkanes conversion: Product identification and effect of cosolvent concentration n-Dodecane and n-tetradecane were tested as MroUPO substrates. The substrate conversion and identification of reaction products were obtained by GC-MS (Table S1). Since the enzymatic oxidation of alkanes in aqueous solutions can be limited by their low solubility in water, these reactions were studied using different acetone:water ratios. The conversion of tetradecane by MroUPO was similar (45 and 50% substrate conversion) in the reactions with 20% and 40% acetone, respectively, and higher than in the reaction with 60% acetone.

194 Annex 7

On the other hand, the relative proportion of the derivatives formed was definitely different (Table S1). The predominance of primary oxidation products (monohydroxylated derivatives) or further oxidized derivatives (aldehydes/ketones, acids, hydroxy/ketone acids) seems related to the higher or lower proportion of acetone in the reaction. With 60% acetone, the monohydroxylated derivatives were predominant (over 80% of total derivatives) after 120 min of reaction, whereas in the reactions with 20% acetone most derivatives (over 80%) were fatty acids and fatty acid derivatives and only a very minor proportion of monohydroxylated derivatives were identified at that reaction time. The reaction with 40% acetone gave similar results as that at 20% although with more presence of monohydroxylated derivatives. The predominance of more oxidized derivatives over the monohydroxylated ones observed in the reactions with 20% acetone, could be due to the higher solubility of the primary oxidation products (monohydroxylated derivatives) in the reaction medium compared with the substrate (alkane) that favored the additional hydroxylation of the alcohols over the alkanes. At higher concentration of acetone in the reaction medium the alkanes become more soluble and both reactions are competing. The above results were attained in the reactions using 0.3 mM substrate. However, higher conversion rates (up to 100%) were attained in reactions using lower substrate concentration (0.1 mM) as shown in Figure S1 for tetradecane.

195 7 Annex

Table S1. Abundance (relative percentage) of the oxygenated derivatives (with alcohol, ketone and acid groups) identified by GC-MS in the reactions of n-dodecane and n-tetradecane (0.3 mM) with M. rotula peroxygenase (1 µM) at 120 min reaction times using different acetone concentrations (20%, 40% and 60%).

Dodecane Tetradecane 20% 20% 40% 60% Alcohols 1-OH 0.8 0.5 - 7.9 2-OH 5.6 2.6 8.4 45.9 3-OH 1.7 1.5 4.5 12.8 4-OH 1.7 - 5.2 9.7 5-OH 1.2 - 4.7 7.3 6-OH 1.7 - 4.5 6.0 Aldehydes/ketones aldehyde - - - - 2-keto 20.3 3.9 17.0 - 3-keto 4.6 2.0 3.8 - 4-keto 2.2 1.1 1.2 - 5-keto 1.6 1.3 1.1 - 6-keto - 1.3 1.4 - 7-keto - 1.5 1.6 - Acids fatty acid 22.6 8.0 24.0 10.4 Hydroxy-ketones 2-OH, 11-keto 10.3 - 7.8 - 2-OH, 10-keto 1.1 - 1.8 - 3-OH, 11-keto 1.5 - 2.0 - Di-keto di-keto - 19.9 - - Hydroxy-acids ω-OH 0.9 1.0 0.6 - (ω-1)-OH 7.5 0.6 1.2 - (ω-2)-OH 1.3 0.3 0.2 - (ω-3)-OH 0.9 - - - (ω-4)-OH 0.2 - - - (ω-5)-OH 0.4 - - - α-OH - 1.7 - - Keto-acids (ω-1)-keto 8.2 35.9 7.7 - (ω-2)-keto - 5.7 0.9 - (ω-3)-keto - 5.0 - - (ω-4)-keto - 3.0 - - Di-acids di-COOH 3.9 3.4 0.4 - Total terminal derivatives 46.7 65.1 26.4 27.4 Conversion rate 68% 45% 50% 22%

196 Annex 7

2-keto (-1)-keto-COOH

COOH

2,13-di-keto 1-OH OH-keto 3-OH 2-OH Alk (C14) (-1)-OH-COOH di-COOH

10 12 14 16 18 20 22 Retention time (min)

Figure S1. GC-MS analysis of MroUPO reaction with tetradecane showing the remaining substrate (Alk, alkane) and the terminal (bold), terminal/subterminal (bold italics), and subterminal carboxylic (COOH), hydroxylated (OH) and keto derivatives. Reactions with

0.1 mM substrate, 40% acetone, 10 mM H2O2 and 1 µM of enzyme, incubated for 120 min.

The presence of the different terminal oxygenation products described above in the alkane reactions with MroUPO contrasts with their complete absence in parallel reactions with AaeUPO, where only subterminal hydroxylated derivatives were found (2- and 3-tetradecanol and 2,12-, 2,13- and 3,12-tetradecanediol in the tetradecane reactions shown in Figure 1) in agreement with previous reports.[3-5]

2.2 Terminal hydroxylations of n-alkanes to dicarboxylic acids: 18O-labeling study An 18O-labeling study, using n-tetradecane and n-dodecane as substrates, and 18 16 either H2 O2 or H2 O2 as enzyme cosubstrate, was performed to investigate the origin of the oxygen incorporated during the oxygenation of n-alkanes. The identification by GC-MS of the different oxidation products at terminal positions in the reaction of n-tetradecane is discussed below, and similar results were obtained in the n-dodecane reactions.

197 7 Annex

2.2.1 Fatty alcohol formation A terminal alcohol was identified for the first time in the reaction of an n-alkane with a peroxygenase. The position of the hydroxyl group was determined by the mass spectra of their trimethylsilyl derivatives, as illustrated in Figure 4A (top) for 1-tetradecanol. This spectrum show a prominent fragment at m/z 271 corresponding + to the loss of a methyl from the trimethylsilyl group [M - CH3] and other characteristic fragments (e.g. at m/z 103).

18 In the reaction using H2 O2, mass spectral analysis of the resulting monohydroxylated alkanes showed that characteristic fragments for the 1- tetradecanol had ~90% shifted from the natural abundance m/z 271 (and m/z 103) found in the unlabeled peroxide reaction to m/z 273 (and m/z 105) (Figure 4A, 18 bottom). Therefore, it was evidenced that one oxygen atom derived from H2 O2 was introduced into the 1-methyl group of n-tetradecane during the first hydroxylation step, as verified by the nearly complete transformation (90%) of the 16O-containing diagnostic fragment ion at m/z 271 to the 18O-containing ion at m/z 18 273 (~10% of the original fragments remained in the H2 O2 reactions due to the 90% 18O isotopic purity of the labeled peroxide used).

2.2.2 Fatty acid formation The fatty acids (lauric and myristic acids) were some of the main products identified in the reactions of MroUPO with dodecane and tetradecane, respectively. 18 18 Incorporation of O from H2 O2 to the carboxyl group was observed during the oxidation of n-tetradecane to myristic acid (Figure 4B). Mass spectral analysis of + the myristic acid formed, showed that the characteristic fragment at [M - CH3] had shifted from the natural abundance m/z 285 found in the unlabeled peroxide reaction (Figure 4B, top) to m/z 289 (incorporation of two 18O atoms at the carboxyl group) (Figure 4B, bottom). Likewise, the characteristic fragment at m/z 117 had shifted to m/z 121 (incorporation of two 18O atoms at the carboxyl group). The small fragment at m/z 300 corresponding to molecular ion also shifted to m/z 304. Finally, 10% unlabeled acid was also formed due to the partial isotopic purity of the peroxide used.

198 Annex 7

2.2.3 Hydroxyfatty acid formation The -hydroxyfatty acids (12-hydroxylauric and 14-hydroxymyristic acids) were identified in the reactions of MroUPO with dodecane and tetradecane, respectively, although they were present in minor amounts.

18 18 The incorporation of O from H2 O2 was evidenced in the mass spectra of the + 14-hydroxymyristic acid (Figure S2A). The characteristic fragments at [M - CH3] had shifted from the natural abundance m/z 373 found in the unlabeled peroxide reaction (Figure S2A, top) to m/z 379 (incorporation of three 18O atoms) (Figure S2A, bottom). Likewise, the fragments at [M - 31]+ and [M - 105]+, resulting from trimethylsilyl and hydrogen transfers during EI mass spectral fragmentation of hydroxycarboxylic acid trimethylsilyl derivatives[6] shifted from the natural abundance m/z 357 and m/z 283, respectively, found in the unlabeled peroxide reaction to m/z 363 and m/z 287.

2.2.4 Dicarboxylic acid formation The dicarboxylic acids (dodecanedioic and tetradecanedioic acids) were identified in the reactions of MroUPO with dodecane and tetradecane, respectively, although in lower amount that the corresponding monocarboxylic acids.

18 18 The incorporation of O from H2 O2 was evidenced in the mass spectra of the + tetradecanedioic acid (Figure S2B). The characteristic fragments at [M - CH3] had shifted from the natural abundance m/z 387 found in the unlabeled peroxide reaction (Figure 4B, top) to m/z 395 (incorporation of four 18O atoms) (Figure S2B, bottom).

199 7 Annex

73 100% 16O A TMS16O 16OTMS

+ [M - CH3] 283 357373 55 147 117 204 103 129 217 171 191 241255 [M+] 388 50 100 150 200 250 300 350 400 m/z

73 100% 18O TMS18O 18OTMS

69 149 97 + 121 287 363[M - CH3] 379 133 161 191 221 206 238254267 50 100 150 200 250 300 350 400 m/z

73 100% 16O TMS16O B 16OTMS 16 + O [M - CH3] 387

117 55 129 204 95 217 271 149 163 170 241 343 371 187 295315329 [M+] 402 50 100 150 200 250 300 350 400 m/z

73 100% 18O TMS18O 18OTMS 18O

+ 121 [M - CH3] 395 95 133 208221 57 149 275 393 172 183 245 299 319 347 50 100 150 200 250 300 350 400 m/z Figure S2. Mass spectra of -hydroxymyristic (A) and tetradecanedioic (B) acids from MroUPO reactions with n-tetradecane in 18O-labeling experiments (bottom) and controls (top).

200 Annex 7

3. Supplemental acknowledgement

René Ullrich and Martin Hofrichter (Technical University of Dresden, Zittau, Germany) are acknowledged for the AaeUPO sample used for comparative purposes.

4. Supplemental references

[1] G. Gröbe, M. Ullrich, M. Pecyna, D. Kapturska, S. Friedrich, M. Hofrichter, K. Scheibner, AMB Express 2011, 1, 31-42. [2] R. Ullrich, J. Nuske, K. Scheibner, J. Spantzel, M. Hofrichter, Appl. Environ. Microbiol. 2004, 70, 4575-4581. [3] A. Gutiérrez, E. D. Babot, R. Ullrich, M. Hofrichter, A. T. Martínez, J. C. del Río, Arch. Biochem. Biophys. 2011, 514, 33-43. [4] E. D. Babot, J. C. del Río, L. Kalum, A. T. Martínez, A. Gutiérrez, Biotechnol. Bioeng. 2013, 110, 2332. [5] S. Peter, M. Kinne, X. Wang, R. Ulrich, G. Kayser, J. T. Groves, M. Hofrichter, FEBS J. 2011, 278, 3667-3675. [6] J. F. Rontani, C. Aubert, J. Amer. Soc. Mass Spectrom. 2008, 19, 66-75.

5. Complete reference cited in the main text

[23] D. Floudas, M. Binder, R. Riley, K. Barry, R. A. Blanchette, B. Henrissat, A. T. Martínez, R. Otillar, J. W. Spatafora, J. S. Yadav, A. Aerts, I. Benoit, A. Boyd, A. Carlson, A. Copeland, P. M. Coutinho, R. P. de Vries, P. Ferreira, K. Findley, B. Foster, J. Gaskell, D. Glotzer, P. Górecki, J. Heitman, C. Hesse, C. Hori, K. Igarashi, J. A. Jurgens, N. Kallen, P. Kersten, A. Kohler, U. Kües, T. K. A. Kumar, A. Kuo, K. LaButti, L. F. Larrondo, E. Lindquist, A. Ling, V. Lombard, S. Lucas, T. Lundell, R. Martin, D. J. McLaughlin, I. Morgenstern, E. Morin, C. Murat, M. Nolan, R. A. Ohm, A. Patyshakuliyeva, A. Rokas, F. J. Ruiz-Dueñas, G. Sabat, A. Salamov, M. Samejima, J. Schmutz, J. C. Slot, F. St.John, J. Stenlid, H. Sun, S. Sun, K. Syed, A. Tsang, A. Wiebenga, D. Young, A. Pisabarro, D. C.

201 7 Annex

Eastwood, F. Martin, D. Cullen, I. V. Grigoriev, D. S. Hibbett, Science 2012, 336, 1715-1719.

202