Study Towards Carotenoid 1,2-Hydratase and as Novel Biocatalysts

Aida HISENI

Study Towards Carotenoid 1,2-Hydratase and Oleate Hydratase as Novel Biocatalysts

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Technische universiteit Delft, op gezag van de Rector Magnificus prof. ir. K.C.A.M Luyben, voorzitter van het College voor promoties, in het openbaar te verdedigen op dinsdag 22 april 2014 om 10:00 uur

door

Aida HISENI

Diplom-Biologin, Heinrich-Heine-Universität Düsseldorf geboren te Doboj, Bosnië en Hercegovina. Dit proefschrift is goedgekeurd door de promotor: Prof. dr. I.W.C.E Arends

Samenstelling promotiecommissie: Rector Magnificus voorzitter Prof. dr. I.W.C.E. Arends Technische Universiteit Delft, promotor Prof. dr. U. Hanefeld Technische Universiteit Delft Prof. dr. J.H. de Winde Universiteit Leiden Prof. dr. G. Muijzer Universiteit van Amsterdam Prof. dr. R. Wever Universiteit van Amsterdam Dr. L.G. Otten Technische Universiteit Delft Dr. P. Dominguez De Maria Sustainable Momentum

Prof. dr. S. de Vries Technische Universiteit Delft, reservelid

This project is financially supported by The Netherlands Ministry of Economic Affairs and the B-Basic partner organizations (http://www.b-basic.nl) through B-Basic, a public- private NWO-ACTS programme [Advanced Chemical Technologies for Sustainability (ACTS)].

ISBN

Copyright © 2014 by Aida HISENI All rights reserved. No part of this publication may be reproduced or distributed in any form or by any means, or stored in a database or retrieval system, without any prior permission of the copyright owner.

To my father Ismet Nukičić

Table of Contents

1 General introduction ...... 1 1.1 and biocatalysis ...... 2 1.2 Enzymes as industrial biocatalysts ...... 3 1.3 engineering ...... 6 1.4 Hydro- ...... 8 1.4.1 Non-enzymatic water addition to a carbon-carbon double bond ...... 8 1.4.2 Enzymatic water addition to a carbon-carbon double bond ...... 9 1.4.3 Carotenoid 1,2-hydratase ...... 11 1.4.4 Oleate hydratase ...... 14 1.5 Scope and objectives ...... 22 1.6 Supplementary figures...... 24 1.7 References ...... 26

2 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina ...... 33 Abstract ...... 34 2.1 Introduction...... 35 2.2 Materials and methods ...... 36 2.2.1 Construction of pET15b_CrtCRg and pET15b_CrtCTr expression vectors ...... 36 2.2.2 Expression and purification of recombinant proteins ...... 37 2.2.3 Tandem MS analysis ...... 37 2.2.4 CrtC activity assay and analysis of the products ...... 38 2.2.5 specificity ...... 38 2.2.6 Effects of pH and temperature on CrtC activity ...... 39 2.2.7 Effects of inhibitors and metal ions on enzyme activity ...... 39 2.2.8 Circular dichroism (CD) spectroscopy ...... 39 2.2.9 Metal analysis using USN-ICP-OES ...... 40 2.3 Results ...... 40 2.3.1 Expression and purification of the carotenoid 1,2-hydratases ...... 40 2.3.2 Hydratase activity ...... 41 2.3.3 ...... 41 2.3.4 Substrate specificity ...... 43 2.3.5 Effect of pH and temperature on hydratase activity and stability ...... 43 2.4 Discussion ...... 46 2.5 Acknowledgments ...... 49 2.6 Supplementary information ...... 50 2.7 References ...... 57

3 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina ...... 59 Abstract ...... 60

vii

3.1 Introduction ...... 61 3.2 Materials and methods ...... 63 3.2.1 In silico analysis ...... 63 3.2.2 Cloning of carotenoid 1,2-hydratase genes ...... 63 3.2.3 Construction of CrtC mutants ...... 64 3.2.3.1 Single point mutations ...... 64 3.2.3.2 N-terminally truncated Rg- and TrCrtC’s ...... 65 3.2.4 Recombinant expression of CrtC’s ...... 65 3.2.5 CrtC purification ...... 66 3.2.6 Determination of enzyme activity ...... 66 3.3 Results and discussion ...... 67 3.3.1 Comparative in silico analysis of crtC genes ...... 67 3.3.2 Production of recombinant wildtype and mutant CrtC’s and enzymatic activity ...... 72 3.4 Conclusion ...... 80 3.5 Acknowledgements ...... 81 3.6 References ...... 82

4 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection ...... 85 Abstract ...... 86 4.1 Introduction ...... 87 4.2 Materials and methods ...... 89 4.2.1 Standard curves and Z-factor determination ...... 89 4.2.2 Large scale production of 10-HSA ...... 90 4.2.3 Growth conditions in 96-well deep well plates ...... 90 4.2.4 Liquid handling ...... 91 4.2.5 Assay conditions ...... 91 4.2.6 Preparation of ohyA mutant libraries ...... 92 4.2.7 Expression of ohyA variants ...... 93 4.2.8 Library screening ...... 93 4.3 Results and discussion ...... 94 4.3.1 Method performance and linearity with small substrates...... 94 4.3.2 Method performance for larger substrates and reaction simulation ...... 95 4.3.3 Precision and accuracy (Z-factor) ...... 9 8 4.3.4 Optimization of protein expression conditions ...... 100 4.4 References ...... 106

5 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA) ...... 109 Abstract ...... 110 5.1 Introduction ...... 111 5.2 Materials and Methods ...... 113 5.2.1 Bacterial strain, growth conditions and cell disruption ...... 113 5.2.2 Precipitation procedure ...... 113 5.2.3 Cross-linking procedure ...... 113 5.2.4 Activity assay...... 114 5.2.5 Storage stability ...... 115 5.2.6 pH activity and temperature stability ...... 115 5.2.7 Biocatalyst recovery ...... 115 viii

5.3 Results and discussion ...... 115 5.3.1 Selection of the best precipitating agent for CLEA preparation ...... 116 5.3.2 Cross-linking and the effect of glutaraldehyde concentration ...... 117 5.3.3 Thermal stability and pH profile of OHase CLEA’s ...... 120 5.3.4 Storage stability of OHase CLEA’s ...... 122 5.3.5 Recycling of OHase CLEA’s ...... 124 5.3.6 Space-time yields ...... 125 5.4 Conclusion ...... 125 5.5 Supplemental Information ...... 127 5.6 References ...... 128

6 Conclusions and future prospects ...... 131 6.1 Carotenoid 1,2-hydratase ...... 132 6.2 Oleate hydratase ...... 134 6.3 High-throughput screening assay ...... 135 6.4 References ...... 138

Summary/Samenvatting ...... 141 Summary ...... 142 Samenvatting ...... 145

Acknowledgements ...... 149

Curriculum vitae ...... 151

ix

Chapter 1 1 General introduction

General introduction

1.1 Enzymes and biocatalysis

Enzymes play a pivotal role in the metabolism of all living organisms. Nearly all biochemical reactions are accomplished and controlled by enzymes. By lowering the activation energy required for a reaction to occur, enzymes are able to dramatically accelerate the reaction rate (up to 1012-fold [1]) for reactions that otherwise would proceed very slowly or not at all. In addition, enzymes are capable of accepting a wide array of complex substrates, are highly selective (enantio-, regio- and chemoselective) and usually operate under mild conditions [2]. Enzymes found in nature have been used since ancient times in the production of food, alcoholic beverages, and manufacturing of commodities such as leather and linen. Jöns Jakob Berzelius, a Swedish chemist, observed in the early nineteenth century that a could be accelerated in the presence of specific compounds. At that time he also coined the term ‘proteins’, without even being aware of the existence of enzymes [3]. Only in the twentieth century, the first enzyme (urease) could be isolated in pure form by James B. Sumner, an American chemist [4]. Since then, enzymes have captured special attention of many researchers. According to Wikipedia, ‘biocatalysis is the use of natural catalysts, such as protein enzymes, to perform chemical transformations on organic compounds. Both enzymes that have been more or less isolated and enzymes still residing inside living cells are employed for this task’. Historically, catalysis is divided into two categories: homogeneous and heterogeneous [5]. Enzymes, however, do not fit into the classical definitions of these two categories. They are usually regarded as a separate class. On the other hand, the development of the biomimetic organocatalysis is causing fading of the boundaries between the catalysis domains. For instance, chemical catalysts are produced, which mimic the natural features of enzymes and also, artificial enzymes are synthesized with specific properties optimized for the targeted application. Despite the early discovery of the catalytic nature of enzymes, their application in industrial processes was not always competitive with chemical catalysis or vice versa [5]. Better understanding of enzyme structure-function relationships and the possibility to tailor their properties have significantly decreased the gap between chemical and enzymatic catalysis [6].

2 1.2 Enzymes as industrial biocatalysts

1.2 Enzymes as industrial biocatalysts

The evolution of modern biotechnology over the last four decades and the emergence of a key technology – genetic engineering - have opened new horizons in the fields of biocatalysis and industrial biotechnology. Today’s novel techniques allow not only the manufacturing of enzymes as purified, well-characterized preparations even on a large scale, but they also make it possible to produce tailor-made enzymes that are designed for a specific and often non-natural application. The application of enzymes as biocatalysts is recognized as a significant complement to the use of chemical reagents. [7] Enzymes are increasingly being utilized for both environmental and economic reasons in a number of industries including agro-food, animal feed, detergent, textile and specialty chemical industry [8, 9] (Table 1.1).

Table 1.1 Enzymes used in various industrial segments and their application, adapted from [10]. Industry Enzyme class Application Detergent Protease Protein stain removal (laundry and dish wash) Amylase Starch stain removal Lipase Lipid stain removal Cellulase Cleaning, color clarification, anti-redeposition (cotton) Mannanase Mannanan stain removal (reappearing stains) Starch and fuel Amylase Starch liquefaction and saccharification Amyloglucosidase Saccharification Pullulanase Saccharification Glucose Glucose to fructose conversion Cyclodextrin- Cyclodextrin production glycosyltransferase Xylanase Viscosity reduction (fuel and starch) Protease Free amino nitrogen production (yeast nutrition - fuel) Food Protease Milk clotting, infant formulas (low allergenic), (including dairy) flavor Lipase Cheese flavor Lactase Lactose removal (milk) Pectin methyl esterase Firming fruit-based products Pectinase Fruit-based products

3 General introduction

Transglutaminase Modify visco-elastic properties Baking Amylase Bread softness and volume, flour adjustment Xylanase Dough conditioning (Phospho)Lipase Dough stability and conditioning (in situ emulsifier) Glucose oxidase Dough strengthening Dough strengthening, bread whitening Protease Biscuits, cookies Transglutaminase Laminated dough strength Animal feed Phytase Phytate digestibility - phosphorus release Xylanase Digestibility β-Glucanase Digestibility Beverage Pectinase De-pectinization, mashing Amylase Juice treatment, low calorie beer β-Glucanase Mashing Acetolactate decarboxylase Maturation (beer) Laccase Clarification (juice), flavor (beer), cork stopper treatment Textile Cellulase Denim finishing, cotton softening Amylase De-sizing Pectate Scouring Catalase Bleach termination Laccase Bleaching Peroxidase Excess dye removal Pulp and paper Lipase Pitch control, contaminant control Protease Biofilm removal Amylase Starch-coating, de-inking, drainage improvement Xylanase Bleach boosting Cellulase De-inking, drainage improvement, fiber modification Fats and oils Lipase Transesterification Phospholipase De-gumming, lyso-lecithin production Organic synthesis Lipase Resolution of chiral alcohols and amines Acylase Synthesis of semi-synthetic penicillin Synthesis of enantiopure carboxylic acids hydratase Synthesis of acrylamide Synthesis of malate Leather Protease Unhairing, bating

4 1.2 Enzymes as industrial biocatalysts

Lipase De-pickling Personal care Amyloglucosidase Antimicrobial (combined with glucose oxidase) Glucose oxidase Bleaching, antimicrobial Peroxidase Antimicrobial

The main benefits offered by enzymes are: (i) waste and energy reduction; enzymes usually work at mild conditions, thereby circumventing the need for harsh chemicals and extreme working conditions; (ii) cleaner products; enzymes are highly specific, resulting in less/no unwanted side reactions and byproducts; (iii) environmental sustainability; enzymes are biodegradable and thereby have no environmental footprint by definition. The synthesis of the drug cortisone is an excellent example of the possibilities of enzyme technology [11, 12]. Here, the number of process steps needed to produce the drug could significantly be reduced from 31 steps (chemical synthesis) to only 5 steps by utilizing enzymes (Figure 1.1).

Figure 1.1 Chemical (upper reaction) and biochemical (lower reaction) route to cortisone (adapted from [5]).

The interest in industrial biocatalysis has increased rapidly and it still continues to grow. It has been estimated that the global market for industrial enzymes is going to reach $6 billion by 2016. Consequently, much effort has been devoted to the development of cleaner alternative technologies where enzymes are utilized as biocatalysts.

5 General introduction

1.3 Enzyme engineering

Despite the huge potential of enzymes in the field of biotechnology, their application is often limited by low stability and/or catalytic activity of these enzymes under process conditions. This is one of the reasons why the use of on industrial scale prevails (Figure 1.2). For instance, lipases which act carboxylic ester bonds, are very versatile enzymes with, among others, broad substrate specificity and stability, and can therefore be utilized in many different industrial applications, such as food, detergent and pharmaceutics [13].

EC 5 () EC 1 EC 4 (Lyases) 2% 4% () 14% EC 2 () 5%

EC 3 (Hydrolases) 75%

Figure 1.2 Pie chart illustrating the utilization of different enzyme classes (EC) on industrial scale, based on Table 1.1.

Therefore, the utilization of enzymes as biocatalysts in industrial processes requires an intensive study and optimization of enzyme properties, such as stability, specific activity and selectivity, beforehand. The development of strategies to overcome the limitations of natural enzymes as biocatalysts has received an enormous boost during the last years [6]. Protein engineering techniques, among others, offer solutions to removing the impediments of widespread application of enzymes in industrial processes. These techniques include random mutagenesis and (semi)rational design/focused mutagenesis (Figure 1.3).

6 1.3 Enzyme engineering

Figure 1.3 Strategies in protein engineering and prerequisites in terms of structural information. Recent methods in diversity generation have been assigned to two categories: (semi)rational design and directed evolution.

The prerequisite for rational design is a detailed structural and mechanistic knowledge of the target enzyme. However, for many enzymes that have been discovered, the X-ray crystal structure or even a viable homology model is not available yet, so that rational mutagenesis of these enzymes is not an option. In contrast, for random mutagenesis, knowledge of the structure-function relationship is not required. In this case, libraries containing a large number of randomly generated mutants with potentially improved and/or novel properties can be produced in a short time. A crucial step in this so-called directed evolution approach is the development of a high- throughput screening (HTS) or a selection assay. The assay allows rapid identification of mutants with the desired properties from a large number of random mutants within a reasonable timeframe [14]. In general, it needs to be sensitive, easy to perform, robust and has to have high throughput. Preferably, screening assays are performed in plate readers, where colorimetric changes or fluorescence formation can be detected upon enzymatic activity. Selection only yields variants that have an advantage over the wild type enzyme in contrast to screens, where the activity of each variant is monitored [15]. In addition, screening methods allow the use of the actual substrate and desired reactions conditions. In the end, “you get what you screened for” [16]).

7 General introduction

1.4 Hydro-Lyases

As indicated above (Figure 1.2), the main EC class that is successfully used in industrial biocatalytic processes is the class of hydrolases (EC 3.-.-.-). Lyases (EC 4.-.-.-), the subject of this thesis, on the other hand, are underrepresented and only a few group members are amenable to be used for industrial scale reactions, including (EC 4.2.1.84) for the production of acrylamide [9] and fumarase (EC 4.2.1.2) to produce malate [17]. Hydro-lyases (EC 4.2.1.-), also called hydratases, are a subclass of carbon-oxygen lyases (EC 4.2.-.-). As a lyase, they catalyze the non-hydrolytic and non-oxidative addition and/or removal of a group to a carbon-carbon double bond. The ‘hydro’ relates to the added or removed group, and is in this case a water molecule.

1.4.1 Non-enzymatic water addition to a carbon-carbon double bond The addition of a water molecule to a non-activated carbon-carbon double bond to yield an alcohol is a very non-selective reaction that requires harsh conditions in traditional chemistry [18]. The non-enzymatic hydration reaction is usually performed by using strong acids, high temperatures and high pressures, or transition metals as a catalyst. Furthermore, the hydration often does not proceed with the desired positional specificity. The acid- catalyzed hydration of an alkene follows the Markovnikov’s rule [19]. It states that the acidic proton binds to the carbon with the greater number of hydrogen atoms, whereas the alcohol group prefers the carbon with the most carbon-carbon bonds (Figure 1.4). The basis of the reaction is the formation of the most stable carbocation, which is subsequently attacked by the nucleophilic water to form the oxonium-ion. Another water molecule takes up the extra proton from the attached oxygen and an alcohol is formed.

Figure 1.4 Acid-catalyzed alkene hydration

8 1.4 Hydro-Lyases

Depending on the structure of the alkene used, unwanted side products and rearrangements can occur, especially with unsymmetrical alkenes. The chemical reaction is therefore limited to alkenes that cannot undergo rearrangement upon hydration.

1.4.2 Enzymatic water addition to a carbon-carbon double bond The enzymatic hydration of carbon-carbon double bonds is catalyzed by hydro-lyases [20]. The alcohol is produced under very mild conditions in a neutral aqueous environment. Due to the inherent high selectivity (enantio-, regio- and chemospecificity) of enzymes, alcohols can be obtained in high yields and without undesired side products. The enzyme database BRENDA [21] counts 153 hydro-lyases (as at December 5, 2013), which catalyze the (de)hydration of a large number of different substrates. Most of these hydro-lyases act on conjugated carbon-carbon double bonds [20, 22]. In contrast to the hydration of isolated carbon-carbon double bond, which is subject to hydronium-ion catalysis (Figure 1.4), a Michael-type hydration occurs for activated double bonds. In this case, the carbon-carbon bond is activated by an electron withdrawing group such as carboxylic acid, thioester or a phosphate group, making it more electrophilic for the nucleophilic addition by water (Figure 1.5).

Figure 1.5 Michael addition of a water molecule to an α,β-unsaturated carbonyl compound.

An excellent overview of these enzymes was recently presented in literature [20]. However, in this thesis, we are aiming for the so far underrepresented class of hydro-lyases acting on isolated carbon-carbon double bonds. Most of the (de)hydratases are dependent (Table 1.2). These cofactors have several functions in the (de)hydration mechanism of (de)hydratases. Next to the direct participation in substrate binding by, for instance, coordination (metal ions, -sulfur clusters), the co-factors can also be involved in the

9 General introduction stabilization of the carbocation intermediate (e.g. pyridoxal phosphate) or are producers of radicals, as found in the dehydration mechanism of diol [23].

Table 1.2 Reported activity requirements for hydratases/. Activity requirement Enzyme name Reference Metal ions Carbonate dehydratase (Zn2+) [24] Phosphopyruvate hydratase (Mg2+) [25] 3-Dehydroshikimate dehydratase (Mn2+) [26] O-succinylbenzoate synthase (Mn2+ or Mg2+) [27] 1,5-Anhydro-D-fructose dehydratase (Ca2+ or Na+ or Mg2+) [28] Coenzyme (Cyanocobalamin) [29] (Pyridoxal 5'-phosphate) [30] Iron-sulfur Aconitate hydratase [31] 2-Methylcitrate dehydratase [32] Methanogen homoaconitase [33] Fumarase (Class I) [34] CoA activated substrates Enoyl-CoA hydratase [35] Long-chain-enoyl-CoA hydratase [36] Heme-thiolate Hydroperoxide dehydratase [37] Colneleate synthase [38] NAD(P)+ / NAD(P)H CDP-glucose 4,6-dehydratase [39] UDP-N-acetylglucosamine 4,6-dehydratase [40] FAD 4-Hydroxybutanoyl-CoA dehydratase [41]

Electron carriers such as NAD+ and NADP+ are usually involved in redox reactions. Therefore, one would not expect to find these cofactors in hydratases (Table 1.2), as the catalytic mechanism of hydratase does not involve a net oxidation or reduction. From a functional point of view, NAD+ behaves as a prosthetic group in hydratases rather than as coenzyme, when it is tightly bound to the enzyme, such as in CDP-glucose 4,6-dehydratase [39]. In this case it initiates the dehydration reaction by oxidation of the substrate and reduction once the substrate has been dehydrated by the enzyme. The catalytic reaction is independent of the NAD+/NADH ratio because of the non-dissociable character of the prosthetic group. The same has been reported for flavoproteins, which catalyze reactions with no net redox change [42]. From an industrial point of view, however, enzymes without the requirement of any cofactor are preferred. The reason is the simplification of the process and no need for the usually expensive cofactors or development of a cofactor regeneration system.

10 1.4 Hydro-Lyases

The following two paragraphs describe two newly discovered hydratases that are cofactor independent and act on isolated carbon-carbon double bonds. Therewith, these are potentially interesting biocatalysts for industrial applications.

1.4.3 Carotenoid 1,2-hydratase Carotenoid 1,2-hydratase (also known as CrtC) is a member of hydro-lyase group EC 4.2.1.131 and occurs in the biosynthetic pathway of carotenoids [43]. From a chemical point of view, CrtC’s are able to perform a very challenging chemical reaction, namely the addition of water to an isolated carbon-carbon double bond [20]. The reaction proceeds with no assistance from electron withdrawing groups, or transition metal cations and does not occur at all under mild conditions in vitro [44]. Carotenoids, which represent one of the most abundant natural pigments with structural and protective properties [45], play an essential role in the photosynthetic machinery of phototrophic organisms such as purple bacteria [46] and higher plants [47]. However, they have also been identified in fungi and some non-photosynthetic bacteria [48]. Depending on the producing organism, carotenoids can be acyclic, monocyclic or bicyclic. CrtC introduces a tertiary hydroxyl group into a carotenoid molecule by addition of water to the carbon-carbon double bond at the C-1 position. The substrate specificity of CrtC’s varies between species. For example, the substrate specificity of the CrtC from Rhodobacter capsulatus is very limited and the enzyme accepts only acyclic carotenoids, which possess two -end groups (acyclic C9 end group according to nomenclature of carotenoids), such as neurosporene and lycopene (Figure 1.6). Once one of the two -end groups is hydrated, the enzyme is not able to use the monohydroxylated carotenoid as a substrate [49]. In contrast, the CrtC’s from Rubrivivax gelatinosus (Rg) and Thiocapsa roseopersicina (Tr) are able to also hydrate monohydroxylated acyclic carotenoids [50]. Next to acyclic carotenoids, the CrtC’s from the purple sulfur bacteria Thiodictyon sp. CAD16 [51] and from the green sulfur bacteria Chlorobium tepidum [52] showed activity towards cyclic carotenoids such as γ-carotene and chlorobactene (Figure 1.6). The substrate specificity of the CrtC and other enzymes involved in the biosynthetic pathway of carotenoids, determine the final structure of accumulated carotenoids in the organism. CrtC belongs to the family PF07143 that encompasses members from several photosynthetic bacteria. Up to now, several carotenoid 1,2-hydratases have been identified in photosynthetic [52-56] as well as in non-photosynthetic bacteria [57, 58]. Recently, carotenoid 1,2-hydratases have been identified in the non-photosynthetic bacterium

11 General introduction

Deinococcus [58], which are able to hydrate γ-carotene, a mono-cyclic substrate, but no acyclic carotenoids.

OH Lycopene

1-HO-Lycopene

Neurosporene OH

1-HO-Neurosporene OH

Demethylspheroidene OCH3

Spheroidene OCH3

1-CH3O-3,4-didehydrolycopene OH

1-HO-3,4-didehydrolycopene

 -Carotene

Chlorobactene

HO Geranylgeraniol (?)

Figure 1.6 Substrates accepted by carotenoid 1,2-hydratases. The shaded portions of each structure are hydrated to yield a tertiary alcohol group. Next to differences in one end group of each carotenoid, the number of double bonds in the molecules differs as well (circled).

They are, however, evolutionary very distinct from the PF07143 members [58] and hence, they have been given the name CruF. Interestingly, cruF homologues are found in a wide variety of carotenoid-synthesizing bacteria that lack a crtC gene [59]. For example, it was found in cyanobacterium Synechococcus sp. [59] and in Herpetosiphon aurantiacus [60]. To our knowledge no published data exist on the catalytic mechanism of CrtC’s, nor has the 3D structure been elucidated yet. However, the 3D structure of the first representative of the Pfam family PF09410 (putative AttH) has been solved, a family which is distantly related to the CrtC family PF07143 [61]. The mechanism of lycopene hydration to hydroxyl compounds, which involves proton attack at C-2 and C-2′ with a carbocation intermediate 2 and the introduction of the hydroxyl group at C-1 and C-1′, was established from H2O- labeling studies with intact cells [62, 63]. Until now no mutagenesis studies have been

12 1.4 Hydro-Lyases published on CrtC, so we identified and mutagenized potential key residues in RgCrtC and TrCrtC. The results of the mutagenesis study together with modeling of a 3D structure with putative AttH led to the hypothesis that the hydration of lycopene is initiated by an acidic residue, Asp268 in RgCrtC and Asp266 in TrCrtC, followed by quenching with solvent water molecules present in the close proximity. From these findings it becomes clear that the complete structure of the enzymes, through crystallization studies, will be pivotal to further unravel the mechanism for this intriguing enzyme. Nevertheless, the results of the study described in chapter 3 shed for the first time light on structure-activity relationships of carotenoid 1,2-hydratases. Whereas CrtC’s from photosynthetic bacteria act on acyclic carotenoids, the CruF’s from non-photosynthetic bacteria only catalyze the hydration of mono-cyclic carotenoids. Protein sequence alignment of CrtC from Rubrivivax gelatinosus and CruF from Deinococcus radiodurans R1 did not reveal any structural similarities (Supplementary figure 1.1). Moreover, they showed substantial differences in the secondary structure. While the CrtC mainly consists of β-strands, the CruF contains notably α-helices (Supplementary figure 1.2). The catalytic and structural features, that determine hydratase activity and specificity of these two distinct families remains hypothetic or unknown. Recently, we recombinantly expressed and characterized two representatives of the PF07143 family, the CrtC from purple non-sulfur Betaproteobacteria Rubrivivax gelatinosus and purple sulfur Gammaproteobacteria Thiocapsa roseopersicina [50]. Biochemical studies revealed that these enzymes are able to convert cofactor independently lycopene into 1-HO-lycopene and 1,1’-(HO)2-lycopene. In addition, they showed some activity towards the unnatural substrate geranylgeraniol, a C20 molecule that resembles the natural substrate lycopene. However, the obtained product could not yet be identified as a hydration product. Furthermore, Steiger et al. [49] have shown that the CrtC from R. gelatinosus also has the ability to hydrate neurosporene, 1-HO-neurosporene and a few other carotenoids [49]. Both CrtC’s are located in the membrane fraction after the heterologous expression in E. coli. The analysis of the amino acid sequence with transmembrane prediction program TMHMM [64] did not reveal any transmembrane segments. However, amino acid region from ca. 120 to 140 is largely hydrophobic in both CrtC’s, which suggest that the enzymes is rather bound to the membrane through an anchor so that a close distance to the substrate, which is synthesized in the cell membranes, is facilitated.

13 General introduction

1.4.4 Oleate hydratase Oleate hydratase (OHase) catalyzes the conversion of oleic acid (OA) into (R)-10- hydroxystearic acid (10-HSA). The enzymatic hydration of OA into 10-HSA (Figure 1.7) was first described in a Pseudomonas strain [65].

Figure 1.7 Conversion of oleic acid into 10-hydroxystearic acid.

Since then reports followed for a series of different bacterial and eukaryotic microorganisms, such as Corynebacterium [66], Saccharomyces cerevisiae [67], Sphingobacterium thalpophilum [68] and Stenotrophomonas nitritireducens [69]. However, no enzyme responsible for this hydration reaction could be identified. Only recently, Bevers et al. [70] succeeded in finding the enzyme and isolating the corresponding gene sequence using the primer walking method. The enzyme was isolated from Elizabethkingia meningoseptica (formerly known as Pseudomonas sp. 3266), the same strain that Davis et al. [71] described 43 years ago. After the recombinant expression in E. coli the enzyme indeed was able to cofactor independently form 10-HSA by hydrating the substrate OA. It is a monomeric 70 kDa soluble enzyme containing one non-essential Ca2+- ion. This hydratase, as well as the carotenoid 1,2-hydratase, represents a new type of hydro- lyase as it is able to hydrate an isolated carbon-carbon double bond. Following its disclosure, OHase has become a favorite topic of many researchers. A number of putative enzymes have been recombinantly expressed, characterized and identified as oleic acid hydratase or fatty acid hydratase. So far, the enzyme has been cloned from Streptococcus pyogenes [72], Bifidobacterium breve [73], Lysinibacillus fusiformi [74, 75], Stenotrophomonas maltophilia [76, 77], Macrococcus caseolyticus [78], Lactobacillus rhamnosus LGG, Lactobacillus plantarum ST-III, Lactobacillus acidophilus NCFM and Bifidobacterium animalis subsp. lactis BB12 [79]. Table 1.3 shows an overview of all characterized OHases and the tested substrates. The results indicate that in all cases, (i) the

14 1.4 Hydro-Lyases carboxylic group, (ii) a minimum distance of nine carbons between the double bond and the acid group, (iii) a minimum chain length of C-14 and (iv) a cis-conformation, are required for conversion of the substrate. For instance, all tested OHases were able to convert the OA into the 10-HSA, while no product was detected when the trans-isomer was used. Furthermore, differences in specificity were observed for the M. casolyticus OHase. In contrast to other OHases, this enzyme introduced a second hydroxyl-group at the C-12 position next to the hydroxyl-group at the C-9 position of the substrate gamma-linolenic acid (C18:3, 6Z, 9Z, 12Z). This could either indicate a true difference in substrate specificity or insufficient incubation time for other OHases, which might have lower reaction rates with this particular substrate. The molecular weight of all reported OHases is ~67 kDa, with B. animalis OHase being the exception with a molecular weight of 82 kDa. OHases from L. fusiformi, S. maltophilia and M. caseolyticus were shown to consist of a dimeric conformational structure upon purification. Although, the OHase from E. meningoseptica is monomeric upon purification, it dimerizes after some time. Furthermore, in our lab it has been established now that the OHase from E. meningoseptica does contain a flavin adenine dinucleotide (FAD) cofactor (Figure 1.8), as well as has been demonstrated for all other oleate hydratases. With the multiple sequences alignment a motif indicative of FAD binding has been identified in all reported and characterized oleic acid hydratases, including that from E. meningoseptica (Figure 1.9). They all share the common conserved sequence motif

GxGxxG(S/A/N)(x)15E(K/D)(x)5E(D/G/S) (where x denotes any amino acid) at the N-terminal part of the sequence, known to bind the FAD cofactor. The first part of the motif (containing the GxGxxG sequence) is the well-known Rossmann fold, a common fold in the FAD- containing glutathione reductase family (GR) [80].

Figure 1.8 Flavin adenine dinucleotide (FAD) cofactor.

15 General introduction

[72] enes g o py Streptococcus Streptococcus

Y N N Y Y Y N N Y

77] [76, maltophilia Stenotrophomonas Stenotrophomonas

Y N Y Y Y Y N Y N

[78] caseolyticus

acrococcus acrococcus M

product detected, -, not Y Y Y Y Y Y - - - Y Y Y

[74,75] ormi f usi f

sinibacillus sinibacillus Ly

Y N Y N Y Y Y Y Y N

[79] LGG rhamnosus

actobacillus actobacillus L Y N Y

. N, no product detected; Y,

[79] ST-III lantarum p

actobacillus actobacillus E. coli L

Y N Y

[79]

acidophilus NCFM NCFM acidophilus

actobacillus actobacillus L

Y N Y

[70] meningoseptica

ia ia g lizabethkin E - Y ------

ses recombinantly expressed expressed ses recombinantly

[73]

idobacterium breve breve idobacterium f i B

Y N Y N Y

[79] B12

oleate hydrata B

animalis subsp. Lactis Lactis subsp. animalis

idobacterium idobacterium f i B ------Y N Y ------

ates tested with stearic acid acid stearic , 9Z, 12Z) - N ------, 11Z, 14Z) , 11Z, 14Z) ------N 10-Hydroxymyristic acid 10,12-Dihydroxy 10-Hydroxyoctadecanoic acid 10-Hydroxyhexadecanoic acid acid 10-Hydroxyhexadecanoic 10,13-Dihydroxyoctadecanoic acid 10-Hydroxy-12(Z)-octadecenoic acid 18:1, 9Z, 12-OH) eic acid (C18:2, 9E, 11E) 9E, 11E) (C18:2, eic acid 9Z, 11E) (C18:2, eic acid ------N N - -

10,13-Dihydroxy-6(Z)-octadecenoic acid 10,13-Dihydroxy-15(Z)-octadecenoic acid 10,13-Dihydroxy-15(Z)-octadecenoic 10-Hydroxy-6(Z),12(Z)-octadecadienoic acid acid 10-Hydroxy-6(Z),12(Z)-octadecadienoic 10-Hydroxy-12(Z),15(Z)-octadecadienoic acid

-Linolenic acid (C18:3,6Z, 9Z, 12Z) -Linolenic(C18:3, acid 9Z, 12Z, 15Z) conjugated- (C18:2, 10E, 12Z) 12Z) 10E, (C18:2, acid conjugated-Linoleic gamma ------N - Linoleic acid methyl ester (C18:2 ester methyl acid Linoleic Stearic acid (C18, no double bond) ------N - - N ------N ------bond) - - 6Z) double no (C18:1, (C18, acid acid bond) double no (C16, acid Palmitic 9Z) (C16:1, acid Palmitoleic Stearic Petroselinic Elaidic acid (C18:1,9E) Oleic acid (C18:1, 9Z) - acid (C Ricinoleic - N - - - - - N - N - N - - N - - N N Vaccenic acid (C18:1, 11Z) (C18:1, 11Z) Vaccenic acid conjugated-Linol conjugated-Linol Linoleic acid (C18:2, 9Z, 12Z) ------N - Arachidic acid (C20, no double bond) ------N - - N ------bond) double no (C20, acid alpha Arachidic Eicosatrienoic acid (C20:3, 3Z, 6Z, 9Z) Arachidonic acid (C20:4, 5Z, 8Z 13Z) Erucic acid (C22:1, 15Z) (C22:1, Nervonic acid Dilinoleoylphosphatidylcholine Trilinoleylglycerol ------N - - N N - - - N N Substrate Substrate Product bond) double no (C12, acid Lauric Myristic acid (C14, no double bond) 9Z) (C14:1, acid Myristoleic ------N N - - Table 1.3 Overview of the substr the of Overview 1.3 Table determined.

16 1.4 Hydro-Lyases

Although distinct conserved sequence motifs were identified in all four FAD families (GR, ferredoxin reductase (FR), p-cresol methylhydroxylase (PCMH) and pyruvate oxidase (PO)), the GxGxxG sequence is the most conserved one and is found in proteins across all four families. The importance of the glycine residues was described by Wierenga et al. [81]. In their study they have been able to derive an amino acid sequence fingerprint, which can be attributed to the so-called βαβ-unit with ADP-binding properties (Figure 1.10). The hydrophobic amino acids of the fingerprint sequence form the hydrophobic core between the β-strand and the α-helix, while the first and the second glycine residues allow a sharp turn and a close approach of the pyrophosphate of the FAD cofactor to the N-terminus of the α-helix, respectively. The acid side-chain at the end of the fingerprint sequence forms a hydrogen bond with the hydroxyl group of the adenine moiety. The amino acid sequence as found in oleate hydratase reveals a slightly different motif compared to the described βαβ-unit with ADP-binding properties (Figure 1.10). It comprises two instead of one acid side-chain. Joo et al. [78] demonstrated by mutagenesis studies the importance of the second acid side-chain in the GxGxxG(S/A/N)(x)15E(K/D)(x)5E(D/G/S) motif (acid side-chain underlined) for the catalytic activity of the oleate hydratase from M. caseolyticus. While a mutant, with the first acid side-chain being substituted by an alanine, retained 60-85% of the wild-type activity, the mutation of the second acid side-chain by an alanine resulted in a fully inactivated enzyme. As already pointed out, the presence of the highly conserved GxGxxG motif in all FAD protein families indicates a crucial role for the molecular recognition of the pyrophosphate moiety. In contrast, residues involved in the binding of the isoalloxazine- and adenine moiety are less conserved and show higher diversity between all the FAD- family members. The isoalloxazine ring structure of the FAD cofactor is involved in the catalytic function in enzymes which catalyze redox reactions. The absent conserved motif for the binding of this part of the FAD molecule within the reported oleate hydratase sequences is consistent with the known fact that the hydration mechanism of these enzyme does not involve any redox reactions [82]. The partly conserved FAD-binding motif and the experimental data on cofactor removal by heat precipitation [78] show that the cofactor in oleate hydratases is held together by weak non-covalent rather than covalent bonds.

17 General introduction

* * *

* *

18 1.4 Hydro-Lyases

Figure 1.9 Multiple sequence alignment showing conserved amino acids of the oleate hydratase (OHase) protein sequences from various bacteria. Identical amino acids are highlighted in black. Sequences analyzed: Elizabethkingia meningoseptica (GI 380877058), Lysinibacillus fusiformis (GI 424736965), Macrococcus caseolyticus (GI 222150326), Lactobacillus acidophilus (GI 58336974), Stenotrophomonas maltophilia (GI 459793677), Streptococcus pyogenes (GI 383479572), Bifidobacterium breve (GI 290048343), Bifidobacterium animalis (GI 384190730), Lactobacillus plantarum (GI 308179305), Lactobacillus rhamnosus (GI 258507498). The predicted FAD binding residues (G70, G72, G75, K91 and E97 of oleate hydratase from Elizabethkingia meningoseptica) are indicated with an asterisk.

Furthermore, through mutagenesis studies of the glycine residues in the oleate hydratase from M. caseolyticus [78] the crucial role for the binding of the cofactor was demonstrated. The molecular interaction of the obtained mutants with FAD was significantly reduced and resulted in inactivation of the enzymatic activity.

Figure 1.10 Schematic drawing of the βαβ-fold from spiny dogfish lactate dehydrogenase with ADP-binding properties, adapted from [81]. The properties of the amino acid residues are indicated with symbols (triangle, hydrophilic or basic; square, hydrophobic and small).

The structural and mechanistic data of oleate hydratases were not available until only recently. Volkov et al. [83] succeeded in determining the crystal structure of oleate hydratase from L. acidophilus, which shares 40% and 57% amino acid sequence identity and similarity, respectively, with that from E. meningoseptica (the enzyme we use in this

19 General introduction research). The enzyme has been crystallized in apo and product-bound (linoleic acid) form and is shown to consist of two stably bound monomers. Upon dimerization, 9.7% of the surface of each monomer becomes buried. Based on the structural similarity to other FAD- binding proteins four different domains were identified (Figure 1.11). Domain 1 (D1) consists of four regions throughout the whole gene and resembles a variant of the Rossmann fold. Domain 2 (D2) is shown to contain the FAD-substrate binding sites in concert with D1. Domain 3 (D3) and domain 4 (D4) comprise only α-helices. The latter was shown to have structural similarity with the N-terminal lid domain of the long-chain acylglycerol lipase from Archaeoglobus fulgidus. Based on the comparison of the obtained apo- and product-bound OHase structures, a displacement of 2 α-helices at the C-terminal region in D4 is observed upon binding of the substrate linoleic acid. This led the authors to the hypothesis that a cavity forming the entrance to a channel is generated, which runs from the surface down to a cleft at the interface of D1 and D3.

D3 D2 D4

D1

Figure 1.11 Crystal structure of Lactobacillus acidophilus hydratase, adapted from [83]. Domains 1, 2 and 3 are shown in marine, light green and red color, respectively. The ‘flexible’ domain 4 is depicted in yellow. Solvent molecules are shown in ball-and-stick representation and are colored cyan. Channel leading to a putative and a putative FAD- are depicted as transparent and yellow surfaces, respectively.

The interior of the channel mainly contains hydrophobic side chains, which accommodate the long fatty acid chain, while positively charged residues at the entrance of the channel (D4) possibly facilitate the recruitment of fatty acids by making a salt bridge to its carboxyl

20 1.4 Hydro-Lyases group. The inability of oleate hydratases to convert substrates lacking the carboxylic group, such as methyl linoleate [72, 73], argue in favor of the substrate recognition at the entrance of the channel. Due to the fact that the crystallization of the hydratase with the bound cofactor FAD was not successful, the FAD-binding domains were only identified through structural similarities to other FAD-binding proteins. Residues involved in the binding of the cofactor are similarly arranged as depicted in Figure 1.10. While the first two glycines are located in the loop region between the first β-strand and α-helix, and help the positioning of the sugar group of the cofactor, the third glycine is positioned in the loop region where the isoalloxazine ring resides. The acid-side chain glutamate (corresponds to E97 in E. meningoseptica OHase), located right after the second β-strand, forms a hydrogen bond to the hydroxyl group of the sugar moiety of the FAD molecule. The here proposed architecture of the FAD binding site is in agreement with the above described. With the availability of the first crystal structure of an oleate hydratase, it was possible to use it as a template in order to generate a model for oleate hydratase from E. meningoseptica. As mentioned earlier, the amino acid similarity of these two hydratases is 57%, which should be sufficient for a reasonable alignment. A model has been generated with an estimated accuracy of 0.63 ± 0.14 (model with a score > 0.5 is considered a good model). Figure 1.12 is a 3D representation of the superimposed structures of oleate hydratases from L. acidophilus and E. meningoseptica and is based on sequence alignment. Overall, both enzymes seem to share similar topology, indicating that these structures are closely related. The topology of the C-terminal region (D4) and D2 region comprising the FAD-binding site, however, has diverged. Volkov et al. [83] proposed the D4 region, which consist of 2 α-helices, as the entrance of the substrate to the active site. In case of E. meningoseptica oleate hydratase one of the two helices is extended, which might indicate different substrate specificity.

21 General introduction

D3 D2 D4

D1

Figure 1.12 Superimposed 3D structures of oleate hydratases from L. acidophilus (cyan) and E. meningoseptica (green), which are based on sequence alignment.

1.5 Scope and objectives

“We do not inherit the earth from our ancestors; we borrow it from our children” (Native American proverb). This quote perfectly pictures the motivation of the research described in this thesis. The environmental change has long-reaching consequences and now, it has been recognized that the development of sustainable and green technologies is of vital importance for our future. As has been introduced in this chapter, enzymes have a great potential in the field of industrial biotechnology. Specifically, hydratases could make very valuable biocatalysts for the chemical industry. The aim of the research described in this thesis was to gain knowledge on structure-function relationship of two newly discovered hydratases, namely carotenoid 1,2-hydratase and oleate hydratase. Based on that, the objective was to map the potential of these two hydratases for their use as biocatalysts in industrial processes. Chapter 2 describes the characterization of carotenoid 1,2- hydratases from photosynthetic bacteria Rubrivivax gelatinosus and Thiocapsa roseopersicina. The biochemical properties

22 1.5 Scope and objectives of the recombinant enzymes and their substrate specificities were studied. In Chapter 3, the two hydratases described in chapter 2 were subjected to protein engineering techniques site-directed evolution and semi-rational mutagenesis in order to identify relevant amino acids in the active site and their contribution to enzymatic activity. Homology modeling together with mutagenesis study helped to gain insight into the enzymatic mechanism of these enzymes. Chapter 4 focuses on the development of a high-throughput screening assay for the detection of alcohols, products of hydrating enzymes such as carotenoid 1,2- hydratase and oleate hydratase (OHase). For this study, OHase from Elizabethkingia meningoseptica was used as the model enzyme. The assay allows for screening of mutant libraries generated by directed evolution. A continuation of the characterization work of OHase that was performed by Loes Bevers, included the study of OHase immobilization as cross-linked enzyme aggregates (CLEA) with the goal to develop OHase into a useful and efficient biocatalyst for high-added value compounds (Chapter 5). In Chapter 6, the main findings described in this thesis are evaluated with the respect to the implications of the work to the fundamental knowledge on carotenoid 1,2-hydratases and oleate hydratases. Next to that, perspectives for future research are presented. Finally, the main findings of this thesis are summarized.

23 General introduction

1.6 Supplementary figures

Supplementary figure 1.1 Sequence alignment and secondary structure prediction (PRALINE software) of carotenoid 1,2-hydratase (CrtC) from Rubrivivax gelatinosus and the evolutionary distinct carotenoid 1,2- hydratase (CruF) from Deinococcus radiodurans R1[46].

24 1.6 Supplementary figures

Supplementary figure 1.2 Sequence alignment and secondary structure prediction (PRALINE software) of carotenoid 1,2-hydratases from Rubrivivax gelatinosus and Thiocapsa roseopersicina.

25 General introduction

1.7 References

[1] E. T. Farinas, et al., "Directed enzyme evolution," Current Opinion in Biotechnology, vol. 12, pp. 545-551, 2001. [2] M. T. Reetz, "Biocatalysis in organic chemistry and biotechnology: Past, present, and future," Journal of the American Chemical Society, vol. 135, pp. 12480-12496, 2013. [3] J. Wisniak, "Jöns Jacob Berzelius A Guide to the Perplexed Chemist," The Chemical Educator, vol. 5, pp. 343-350, 2000/12/01 2000. [4] J. B. Sumner, "The isolation and crystallization of the enzyme urease: Preliminary paper," Journal of Biological Chemistry, vol. 69, pp. 435-441, August 1, 1926 1926. [5] R. Yuryev and A. Liese, "Biocatalysis: The Outcast," ChemCatChem, vol. 2, pp. 103-107, 2010. [6] U. T. Bornscheuer, et al., "Engineering the third wave of biocatalysis," Nature, vol. 485, pp. 185- 194, 2012. [7] N. Ran, et al., "Recent applications of biocatalysis in developing green chemistry for chemical synthesis at the industrial scale," Green Chemistry, vol. 10, pp. 361-372, 2008. [8] S. S. Dewan, "Enzymes in Industrial Applications: Global Markets," Report BIO030G, BCC Research, Wellesley, MD, USA, 2012. [9] K. R. Jegannathan and P. H. Nielsen, "Environmental assessment of enzyme use in industrial production-a literature review," Journal of Cleaner Production, vol. 42, pp. 228-240, 2013. [10] O. Kirk, et al., "Industrial enzyme applications," Current Opinion in Biotechnology, vol. 13, pp. 345-351, 2002. [11] L. H. Sarett, "Partial synthesis of pregnene-4-triol-17(β),20(β),21-dione-3,11 and pregnene-4-diol- 17(β),21-trione-3,11,20 monoacetate," Journal of Biological Chemistry, vol. 162, pp. 601-631, March 1, 1946 1946. [12] O. K. Sebek and D. Perlman, "Chapter 14 - Microbial transformation of steroids and sterols," in Microbial Technology (Second Edition), H. J. Peppler and D. Perlman, Eds., ed: Academic Press, 1979, pp. 483-496. [13] A. Houde, et al., "Lipases and their industrial applications: An overview," Applied Biochemistry and Biotechnology - Part A Enzyme Engineering and Biotechnology, vol. 118, pp. 155-170, 2004. [14] J.-L. Reymond, Enzyme assays: High-throughput screening, genetic selection and fingerprinting: Whiley-VCH, 2006. [15] M. Goldsmith and D. S. Tawfik, "Directed enzyme evolution: Beyond the low-hanging fruit," Current Opinion in Structural Biology, vol. 22, pp. 406-412, 2012. [16] L. You and F. H. Arnold, "Directed evolution of subtilisin E in Bacillus subtilis to enhance total activity in aqueous dimethylformamide," Protein Engineering, vol. 9, pp. 77-83, 1996. [17] A. V. Presečki and D. Vasić-Rački, "Production of L-malic acid by permeabilized cells of commercial Saccharomyces sp. strains," Biotechnology Letters, vol. 27, pp. 1835-1839, 2005.

26 1.7 References

[18] M. L. Bender and K. A. Connors, "A non-enzymatic olefinic hydration under neutral conditions. The kinetics of the hydration of fumaric acid monoanion," Journal of the American Chemical Society, vol. 83, pp. 4099-4100, 1961. [19] G. Jones, "The Markovnikov rule," Journal of Chemical Education, vol. 38, p. 297, 1961/06/01 1961. [20] J. F. Jin and U. Hanefeld, "The selective addition of water to C=C bonds; enzymes are the best chemists," Chemical Communications, vol. 47, pp. 2502-2510, 2011. [21] I. Schomburg, et al., "BRENDA in 2013: integrated reactions, kinetic data, enzyme function data, improved disease classification: new options and contents in BRENDA," Nucleic Acids Research, vol. 41, pp. D764-D772, January 1, 2013 2013. [22] C. Wuensch, et al., "Asymmetric enzymatic hydration of hydroxystyrene derivatives," Angewandte Chemie - International Edition, vol. 52, pp. 2293-2297, 2013. [23] T. Toraya, "Radical catalysis in coenzyme B12-dependent isomerization (eliminating) reactions," Chemical Reviews, vol. 103, pp. 2095-2128, 2003/06/01 2003.

[24] X. Zhang, et al., "Kinetics and mechanism of the hydration of CO2 and dehydration of HCO3 - catalyzed by a Zn(II) complex of 1,5,9-triazacyclododecane as a model for ," Inorganic Chemistry, vol. 32, pp. 5749-5755, 1993. [25] N. Saito, "Purification and properties of bacterial phosphopyruvate hydratase," Journal of Biochemistry, vol. 61, pp. 59-69, 1967. [26] B. F. Pfleger, et al., "Structural and functional analysis of AsbF: Origin of the stealth 3,4- dihydroxybenzoic acid subunit for petrobactin biosynthesis," Proceedings of the National Academy of Sciences of the United States of America, vol. 105, pp. 17133-17138, 2008. [27] T. B. Thompson, et al., "Evolution of enzymatic activity in the superfamily: Structure of o- succinylbenzoate synthase from Escherichia coli in complex with Mg2+ and o-succinylbenzoate," Biochemistry, vol. 39, pp. 10662-10676, 2000/09/01 2000. [28] A. J. Morgan, et al., "1, 5-Anhydro-D-fructose dehydratase," ed: Google Patents, 2002. [29] T. Toraya, et al., "Propanediol dehydratase system. Role of monovalent cations in binding of vitamin B12 coenzyme or its analogs to apoenzyme," Biochemistry, vol. 10, pp. 3475-3484, 1971. [30] P. Bartholmes, et al., "Cooperative and noncooperative binding of pyridoxal 5′-phosphate to tryptophan synthase from Escherichia coli," Biochemistry, vol. 15, pp. 4712-4717, 1976. [31] M. M. Werst, et al., "Characterization of the [4Fe-4S]+ cluster at the active site of by 57Fe, 33S, and 14N electron nuclear double resonance spectroscopy," Biochemistry, vol. 29, pp. 10533- 10540, 1990. [32] T. L. Grimek and J. C. Escalante-Semerena, "The acnD genes of Shewenella oneidensis and Vibrio cholerae encode a new Fe/S-dependent 2-methylcitrate dehydratase enzyme that requires prpF function in vivo," Journal of Bacteriology, vol. 186, pp. 454-462, 2004. [33] R. M. Drevland, et al., "Methanogen homoaconitase catalyzes both hydrolyase reactions in coenzyme B biosynthesis," Journal of Biological Chemistry, vol. 283, pp. 28888-28896, 2008. [34] B. M. A. van Vugt-Lussenburg, et al., "Identification of two [4Fe-4S]-cluster-containing hydro- lyases from Pyrococcus furiosus," Microbiology, vol. 155, pp. 3015-3020, 2009.

27 General introduction

[35] H. A. Hofstein, et al., "Role of glutamate 144 and glutamate 164 in the catalytic mechanism of enoyl- CoA hydratase," Biochemistry, vol. 38, pp. 9508-9516, 1999/07/01 1999. [36] J. C. Fong and H. Schulz, "Purification and properties of pig heart crotonase and the presence of short chain and long chain enoyl coenzyme A hydratases in pig and guinea pig tissues," Journal of Biological Chemistry, vol. 252, pp. 542-547, 1977. [37] Y. Shibata, et al., "Fatty acid is a heme protein," Biochemical and Biophysical Research Communications, vol. 207, pp. 438-443, 1995. [38] A. N. Grechkin, "Hydroperoxide lyase and divinyl ether synthase," Prostaglandins and Other Lipid Mediators, vol. 68-69, pp. 457-470, 2002. [39] X. He, et al., "Probing the coenzyme and substrate binding events of CDP-D-glucose 4,6- dehydratase: Mechanistic implications," Biochemistry, vol. 35, pp. 4721-4731, 1996. [40] J. P. Morrison, et al., "Mechanistic studies on PseB of pseudaminic acid biosynthesis: A UDP-N- acetylglucosamine 5-inverting 4,6-dehydratase," Bioorganic Chemistry, vol. 36, pp. 312-320, 2008. [41] U. Müh, et al., "4-Hydroxybutyryl-CoA dehydratase from Clostridium aminobutyricum: Characterization of FAD and iron−sulfur clusters involved in an overall non-redox reaction," Biochemistry, vol. 35, pp. 11710-11718, 1996/01/01 1996. [42] S. Bornemann, "Flavoenzymes that catalyse reactions with no net redox change," Natural Product Reports, vol. 19, pp. 761-772, 2002. [43] D. Umeno, et al., "Diversifying carotenoid biosynthetic pathways by directed evolution," Microbiology and Molecular Biology Reviews, vol. 69, pp. 51-78, Mar 2005. [44] C. M. Evans and A. J. Kirby, "A model for olefin hydration: Intramolecular nucleophilic addition of phenolate oxygen to the unactivated double bond," Journal of the Chemical Society, Perkin Transactions 2, pp. 1259-1267, 1984. [45] G. A. Armstrong and J. E. Hearst, "Carotenoids .2. Genetics and molecular biology of carotenoid pigment biosynthesis," Faseb Journal, vol. 10, pp. 228-237, Feb 1996. [46] S. L. Jensen, et al., "Biosynthesis of carotenoids in purple bacteria : A re-evaluation based on considerations of chemical structure," Nature, vol. 192, pp. 1168-1172, 1961. [47] C. I. Cazzonelli, "Carotenoids in nature: insights from plants and beyond," Functional Plant Biology, vol. 38, pp. 833-847, 2011. [48] G. A. Armstrong, "Genetics of eubacterial carotenoid biosynthesis: A colorful tale," Annual Review of Microbiology, vol. 51, pp. 629-659, 1997. [49] S. Steiger, et al., "Heterologous expression, purification, and enzymatic characterization of the acyclic carotenoid 1,2-hydratase from Rubrivivax gelatinosus," Archives of Biochemistry and Biophysics, vol. 414, pp. 51-58, Jun 1 2003. [50] A. Hiseni, et al., "Biochemical characterization of the carotenoid 1,2-hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina," Applied Microbiology and Biotechnology, vol. 91, pp. 1029-1036, 2011. [51] K. Vogl and D. A. Bryant, "Elucidation of the biosynthetic pathway for okenone in Thiodictyon sp. CAD16 leads to the discovery of two novel carotene ketolases," Journal of Biological Chemistry, vol. 286, pp. 38521-38532, November 4, 2011 2011.

28 1.7 References

[52] N. U. Frigaard, et al., "Genetic manipulation of carotenoid biosynthesis in the green sulfur bacterium Chlorobium tepidum," Journal of Bacteriology, vol. 186, pp. 5210-5220, Aug 2004. [53] G. A. Armstrong, et al., "Nucleotide-sequence, organization, and nature of the protein products of the carotenoid biosynthesis gene-cluster of Rhodobacter-capsulatus," Molecular & General Genetics, vol. 216, pp. 254-268, Apr 1989. [54] H. P. Lang, et al., "Complete DNA-sequence, specific Tn5 insertion map, and gene assignment of the carotenoid biosynthesis pathway of Rhodobacter-sphaeroides," Journal of Bacteriology, vol. 177, pp. 2064-2073, Apr 1995. [55] A. T. Kovacs, et al., "Genes involved in the biosynthesis of photosynthetic pigments in the purple sulfur photosynthetic bacterium Thiocapsa roseopersicina," Applied and Environmental Microbiology, vol. 69, pp. 3093-3102, Jun 2003. [56] E. Giraud, et al., "Two distinct crt gene clusters for two different functional classes of carotenoid in Bradyrhizobium," Journal of Biological Chemistry, vol. 279, pp. 15076-15083, Apr 9 2004. [57] J. A. Botella, et al., "A cluster of structural and regulatory genes for light-iduced carotenogenesis in Myxococcus-xanthus," European Journal of Biochemistry, vol. 233, pp. 238-248, Oct 1995. [58] Z. T. Sun, et al., "A novel carotenoid 1,2-hydratase (CruF) from two species of the non- photosynthetic bacterium Deinococcus," Microbiology-Sgm, vol. 155, pp. 2775-2783, Aug 2009. [59] J. E. Graham and D. A. Bryant, "The biosynthetic pathway for myxol-2′ fucoside (myxoxanthophyll) in the cyanobacterium Synechococcus sp. strain PCC 7002," Journal of Bacteriology, vol. 191, pp. 3292-3300, 2009. [60] H. Kiss, et al., "Complete genome sequence of the filamentous gliding predatory bacterium Herpetosiphon aurantiacus type strain (114-95 T)," Standards in Genomic Sciences, vol. 5, pp. 356- 370, 2011. [61] H. J. Chiu, et al., "Structure of the first representative of Pfam family PF09410 (DUF2006) reveals a structural signature of the calycin superfamily that suggests a role in lipid metabolism," Acta Crystallographica Section F: Structural Biology and Crystallization Communications, vol. 66, pp. 1153-1159, 2010. [62] N. J. Patel, et al., "Use of deuterium labelling from deuterium oxide to demonstrate carotenoid transformations in photosynthetic bacteria," BBA - General Subjects, vol. 760, pp. 92-96, 1983. [63] A. A. Yeliseev and S. Kaplan, "Anaerobic carotenoid biosynthesis in Rhodobacter sphaeroides

2.4.1: H2O is a source of oxygen for the 1-methoxy group of spheroidene but not for the 2-oxo group of spheroidenone," Febs Letters, vol. 403, pp. 10-14, 1997. [64] E. L. Sonnhammer, et al., "A hidden Markov model for predicting transmembrane helices in protein sequences," Proceedings / ... International Conference on Intelligent Systems for Molecular Biology ; ISMB. International Conference on Intelligent Systems for Molecular Biology, vol. 6, pp. 175-182, 1998. [65] L. L. Wallen, et al., "The microbiological production of 10-hydroxystearic acid from oleic acid," Archives of Biochemistry and Biophysics, vol. 99, pp. 249-253, 1962. [66] C. W. Seo, et al., "Hydration of squalene and oleic-acid by Corynebacterium sp. S-401," Agricultural and Biological Chemistry, vol. 45, pp. 2025-2030, 1981.

29 General introduction

[67] S. H. Elsharkawy, et al., "Microbial oxidation of oleic-acid," Applied and Environmental Microbiology, vol. 58, pp. 2116-2122, Jul 1992. [68] T. M. Kuo and W. E. Levinson, "Biocatalytic production of 10-hydroxystearic acid, 10-ketostearic acid, and their primary fatty amides," Journal of the American Oil Chemists Society, vol. 83, pp. 671-675, Aug 2006. [69] B. N. Kim, et al., "Conversion of oleic acid to 10-hydroxystearic acid by whole cells of Stenotrophomonas nitritireducens," Biotechnology Letters, vol. 33, pp. 993-997, May 2010. [70] L. E. Bevers, et al., "Oleate hydratase catalyzes the hydration of a nonactivated carbon-carbon bond," Journal of Bacteriology, vol. 191, pp. 5010-5012, Aug 2009. [71] E. N. Davis, et al., "Microbial hydration of cis-9-alkenoic acids " Lipids, vol. 4, pp. 356-362, 1969. [72] A. Volkov, et al., "Myosin cross-reactive antigen of Streptococcus pyogenes M49 encodes a fatty acid double bond hydratase that plays a role in oleic acid detoxification and bacterial virulence," Journal of Biological Chemistry, vol. 285, pp. 10353-10361, 2010. [73] E. Rosberg-Cody, et al., "Myosin-cross-reactive antigen (MCRA) protein from Bifidobacterium breve is a FAD-dependent fatty acid hydratase which has a function in stress protection," Bmc Biochemistry, vol. 12, 2011. [74] B. N. Kim, et al., "Production of 10-hydroxystearic acid from oleic acid and olive oil hydrolyzate by an oleate hydratase from Lysinibacillus fusiformis," Applied Microbiology and Biotechnology, vol. 95, pp. 929-937, 2012. [75] M.-H. Seo, et al., "Production of a novel compound, 10,12-dihydroxystearic acid from ricinoleic acid by an oleate hydratase from Lysinibacillus fusiformis," Applied Microbiology and Biotechnology, pp. 1-9, 2013/02/03 2013. [76] Y. C. Joo, et al., "Production of 10-hydroxystearic acid from oleic acid by whole cells of recombinant Escherichia coli containing oleate hydratase from Stenotrophomonas maltophilia," Journal of Biotechnology, vol. 158, pp. 17-23, 2012. [77] E. Y. Jeon, et al., "Bioprocess engineering to produce 10-hydroxystearic acid from oleic acid by recombinant Escherichia coli expressing the oleate hydratase gene of Stenotrophomonas maltophilia," Process Biochemistry, vol. 47, pp. 941-947, 2012. [78] Y. C. Joo, et al., "Biochemical characterization and FAD-binding analysis of oleate hydratase from Macrococcus caseolyticus," Biochimie, vol. 94, pp. 907-915, 2012. [79] B. Yang, et al., "Myosin-cross-reactive antigens from four different lactic acid bacteria are fatty acid hydratases," Biotechnology Letters, vol. 35, pp. 75-81, 2013. [80] O. Dym and D. Eisenberg, "Sequence-structure analysis of FAD-containing proteins," Protein Science, vol. 10, pp. 1712-1728, 2001. [81] R. K. Wierenga, et al., "Prediction of the occurrence of the ADP-binding βαβ-fold in proteins, using an amino acid sequence fingerprint," Journal of Molecular Biology, vol. 187, pp. 101-107, 1986. [82] W. G. Niehaus Jr, et al., "Stereospecific hydration of the delta-9 double bond of oleic acid," Journal of Biological Chemistry, vol. 245, pp. 3790-3797, 1970.

30 1.7 References

[83] A. Volkov, et al., "Crystal structure analysis of a fatty acid double-bond hydratase from Lactobacillus acidophilus," Acta Crystallographica Section D: Biological Crystallography, vol. 69, pp. 648-657, 2013.

31

Chapter 2

2 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

Aida Hiseni, Isabel W.C.E. Arends and Linda G. Otten

Appl Microbiol Biotechnol. Aug 2011; 91(4): 1029–1036

Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina Abstract

Two carotenoid 1,2-hydratase (crtC) genes from the photosynthetic bacteria Rubrivivax gelatinosus and Thiocapsa roseopersicina were cloned and expressed in Escherichia coli in an active form and purified by affinity chromatography. The biochemical properties of the recombinant enzymes and their substrate specificities were studied. The purified CrtC’s catalyze cofactor independently the conversion of lycopene to 1-HO- and 1,1′-(HO)2- lycopene. The optimal pH and temperature for hydratase activity was 8.0 and 30ºC, respectively. The apparent Km and Vmax values obtained for the hydration of lycopene were 24 µM and 0.31 nmol h-1 mg-1 for RgCrtC and 9.5 µM and 0.15 nmol h-1 mg-1 for TrCrtC, respectively. SDS-PAGE analysis revealed two protein bands of 44kDa and 38kDa for TrCrtC, which indicate protein processing. Both hydratases are also able to convert the unnatural substrate geranylgeraniol (C20 substrate), which functionally resembles the natural substrate lycopene.

34 2.1 Introduction

2.1 Introduction

Optically pure tertiary alcohols are highly valuable building blocks for the synthesis of several bioactive natural products and pharmaceuticals [1]. However, the synthesis of optically pure tertiary alcohols in high yield without undesired side products is still a challenging task in traditional chemical synthesis [2]. Much effort has therefore been devoted to the development of cleaner alternative technologies. The application of biocatalysts is recognized as a significant complement to the use of chemical reagents. Biocatalysts such as enzymes and whole microbial cells are increasingly being utilized for both environmental and economic reasons in a number of industries including agro-food, animal feed, detergent, textile and specialty chemical industry. The market for enzymes has increased in an almost exponential manner from 1960s to 2000 [3]. This is due to the well- known benefits of enzymes. They are remarkable catalysts capable of accepting a wide array of complex substrates, are highly selective (enantio-, regio- and chemoselective) and operate efficiently under mild conditions. The possibility of using enzymes for the production of tertiary alcohols has generated our interest in the enzyme class of hydro-lyases (EC 4.2.1-), which catalyze the reversible addition of water to a carbon-carbon double bond. Although more than 100 hydro-lyases have been discovered to date, only a few examples have been used in industrial applications [4, 5]. For example, for the production of R-γ-dodeca-lactone, an essential flavor in whisky, oleate hydratase has been utilized, which catalyzes the conversion of oleic acid to form (R)- γ-hydroxy-stearate, which again is converted to the end product by baker’s yeast [6-8]. Carotenoid 1,2-hydratase (CrtC), another member of the hydro-lyases group, occurs in the biosynthetic pathway of different acyclic carotenoids in photosynthetic bacteria. CrtC introduces a tertiary hydroxy group into a carotenoid molecule by addition of water to the carbon-carbon double bond at the C-1 position. Several carotenoid 1,2-hydratases have been identified in photosynthetic [9-13] as well as in non-photosynthetic bacteria [14, 15]. Recently, a novel carotenoid 1,2-hydratase (CruF) has been described in the non- photosynthetic bacterium Deinococcus [15], which catalyzes C-1′,2′ hydration of γ- carotene. This enzyme though, is evolutionarily distinct from the CrtC family in photosynthetic bacteria. The CrtC from the purple non-sulfur photosynthetic bacterium Rubrivivax gelatinosus has been partially characterized and it was found to be a membrane-bound enzyme with a

35 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

molecular weight of 44 kDa [16]. In vitro assay showed that the enzyme was able to hydrate the carbon-carbon double bond at the ψ-end group of several natural substrates such as

neurosporene and lycopene to the corresponding products 1-HO- and 1,1′-(HO)2-

neurosporene and 1-HO- and 1,1′-(HO)2-lycopene without the use of any cofactor. Through genetic analysis and characterization of the pigment biosynthesis genes in the purple sulfur photosynthetic bacterium Thiocapsa roseopersicina a putative protein was found that showed high identity to CrtC from R. gelatinosus [11]. Gene cluster analysis of T. roseopersicina (Gammaproteobacteria) revealed a significant identity (55 %) of the crtC gene product to the CrtC from R. gelatinosus (Betaproteobacteria), although the arrangement of the pigment biosynthesis gene cluster resembles more that of Rhodobacter species (Alphaproteobacteria) [17]. However, so far the enzyme has not been isolated or characterized in any detail, which makes it a potential candidate for a hydro-lyase with new properties. In order to make this group of enzymes more attractive for green hydration reactions in industrial applications, we have investigated parameters that could be of major importance to that field. Herein we report on the detailed biochemical characterization of the two CrtC’s from R. gelatinosus and T. roseopersicina. This provides an insight into their potential to be used as biocatalysts. The broad stability and activity profiles of both enzymes are promising for industrial biocatalysis.

2.2 Materials and methods

2.2.1 Construction of pET15b_CrtCRg and pET15b_CrtCTr expression vectors

The crtCRg and crtCTr genes were amplified with primers Rg_fw/Rg_rv (GGGAGTACCATATGCGAGCAGCGGAGTC and ATACACTCGAGATGTATACG TCAAGCGCGG) and Tr_fw/Tr_rv (GGAGTAATCATATGCGAGCAGCGGGC and CCCTCGAGAACTATGTCTTCT-CAGCCGCC), respectively, containing restriction sites for NdeI (forward) and XhoI (reverse) (restriction sites are underlined). Amplification

reactions were done under standard PCR conditions using plasmids pPQE30crtCRg and pTcrt3 respectively, as template (Supplementary table 2.1). Using NdeI/XhoI restriction sites the digested and purified fragment was ligated into the same sites of the pET15b vector

36 2.2 Materials and methods

and transformed into E. coli TOP10 competent cells. The insertion of the crtC gene was verified by restriction analysis with NdeI/XhoI enzymes and DNA sequencing (BaseClear).

2.2.2 Expression and purification of recombinant proteins E. coli BL21 (DE3) was the host for the pET15_CrtC plasmids. Cultures were grown at -1 37°C in LB-broth with 100 µg ml ampicillin until an OD600 value of 0.6-0.8 was reached. Protein expression was induced by addition of isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 0.1 mM, followed by cultivation at 25°C overnight. The induced cells were harvested by centrifugation at 10.000 rpm for 10 min at 4°C (Sorvall), washed once with 50 mM Na2HPO4 buffer (pH 8.0) and suspended in the binding buffer

(50 mM Na2HPO4, 300 mM NaCl, 20 mM imidazole, pH 8.0). Cell-free extract (CFE) was obtained after lysis of the cells with 1 mg ml-1 lysozyme for 1 h at 4ºC followed by cell disruption at the pressure of 2.4 kBar (Constant systems, IUL instruments) and centrifugation at 10.000 rpm for 20 min at 4°C. The separation of the CFE into membrane fraction and supernatant was done by centrifugation at 45.000 rpm for 1 h at 4ºC. CFE’s were filtered through 0.45 µm filter (Whatman, FP 30/0, 45 CA-S) and each extract was applied separately onto Ni-NTA HisTrap HP column (1.6 x 2.5 cm, 5 ml, GE Healthcare) previously equilibrated with binding buffer. The purification and the loading of the samples onto the column were performed with the HPLC-system in conjunction with the LCsolution software (Shimadzu). Unbound proteins were washed from the column with

a gradient of 50-75 mM imidazole in washing buffer (50 mM Na2HPO4, 300 mM NaCl, pH 8.0). Then, the CrtC protein was eluted from the column with a gradient of 75-300 mM

imidazole in elution buffer (50 mM Na2HPO4, 300 mM NaCl, pH 8.0). Enzyme fractions were separated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE; 10% Bis-Tris, BioRad) and visualized by staining with SimplyBlue SafeStain (Invitrogen). Fractions containing CrtC were combined and concentrated using Amicon Ultra-30 filters (Millipore). The concentrated sample was applied onto a PD-10 desalting column (GE

Healthcare) previously equilibrated with 50 mM Na2HPO4 buffer (pH 8.0). The eluted enzyme sample was frozen in liquid nitrogen and stored in aliquots at -80ºC.

2.2.3 Tandem MS analysis The concentrated CrtC sample was further purified by SDS-PAGE. The protein band was excised from the gel and subjected to in-gel proteolytic digestion as previously described [18].

37 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina 2.2.4 CrtC activity assay and analysis of the products Enzyme activity was determined either with the purified enzyme or with the CFE. The

assay was performed with 50-100 µg enzyme in 200 µl 50 mM Na2HPO4 buffer (pH 8.0), containing 10 mg ml-1 L-α-phosphatidylcholine (from egg yolk) and 20 µM lycopene (Lanospharma Laboratories Co., Ltd) from a stock in acetone. After incubation at 28ºC and shaking at 800 rpm in the dark the substrate and products were extracted from the aqueous layer after a desired time interval. Prior to the extraction 50 µl of saturated NaCl solution was added and the carotenoids extracted with one volume of dichloromethane. The mixtures were shaken for 5 min at 1400 rpm, centrifuged for 1 min at 13.200 rpm (Eppendorf) and 150 µl of the dichloromethane phase was dried with a Speed Vac Concentrator (Thermo). The dried carotenoids were dissolved in 10 µl dichloromethane, diluted 1:10 with 100% acetonitrile and analyzed with HPLC. Separation was performed with a Merck 4.6x50 mm Chromolith TM SpeedROD RP-18e 2µm-column with acetonitrile / water (95:5, v/v) as the eluent. Lycopene and the corresponding products were detected at 470 nm (SPD-20A, Shimadzu). Marker carotenoids were obtained as described by Steiger et al. [19] and used for the identification of the reaction products. The lycopene concentration in the assay was quantified from the calibration curve constructed by diluting a stock of lycopene in dichloromethane with acetone. A second calibration curve, which was used to quantify the reaction products, was constructed in the same way as the standard assay, including the extraction step. For the determination of enzyme kinetic parameters, the purified enzyme was incubated for

4 hours with different concentrations of lycopene (0.5-35 µM) in 50 mM Na2HPO4 buffer (pH 8.0), containing 10 mg ml-1 L-α-phosphatidylcholine. Each reaction was performed in

duplicate. The affinity constant (Km) and the maximal velocity (Vmax) were calculated from the experimental data points using OriginPro 8 SR1 software.

2.2.5 Substrate specificity Substrate specificity was assayed using the following acyclic alkenes: 2-methyl-2-butene (79 mM), 2-methyl-2-pentene (68 mM), farnesol (33 mM) and geranylgeraniol (14.3 mM), as substrates. Reactions were carried out using standard assay conditions. E. coli carrying the empty pET15-b vector was used as negative control reaction. Substrates and products were extracted from aqueous layer with one volume of ethyl acetate. The samples were

dried with Na2SO4 prior to their injection. Separation and identification of the components

38 2.2 Materials and methods

was effected with a Shimadzu GC-MS coupled to a QP-2010S with a FactorFour VF- WAXms column (length 30 m, diameter 0.25 mm, and film thickness 0.25 µm).

2.2.6 Effects of pH and temperature on CrtC activity In order to investigate the pH effect on the CrtC activity, the reactions were carried out in buffers with varying pH values. The buffers used for pH test were sodium acetate (100 mM, pH 3.0-6.0), potassium phosphate (100 mM, pH 6.0-8.0) and Tris-HCl (50 mM, pH 8.0- 9.0). The measurements were conducted at 28ºC and lycopene (20 µM) was used as substrate, as described in the section “CrtC activity assay and analysis of the products”. The pH stability of the enzyme was performed by measuring the remaining activity at pH 8.0 after the enzyme had been incubated in the corresponding buffers for 30 min. The optimum temperature for CrtC activity was determined by testing enzyme activity at temperatures ranging from 0 to 50ºC using the standard activity assay. The thermal stability was investigated by pre-incubating the enzyme at various temperatures (5-50ºC) in the absence of substrate for 30 min, cooling the enzyme on ice, and then measuring the residual activity in a standard assay with lycopene as substrate.

2.2.7 Effects of inhibitors and metal ions on enzyme activity The inhibitory effects on enzyme activity were investigated by performing activity assay under standard conditions in the presence of several metal ions (MgCl2, MnCl2, CoCl2, + ZnCl2, CaCl2 and CuSO4) and chemicals (NAD , NADH, protease inhibitor “Complete” (Roche)) with a final concentration of 1 mM. Lycopene (20 µM) was used as substrate and the activity was measured as described above. Reaction mixture without any additive was used as control reaction and was designated as 100% activity.

2.2.8 Circular dichroism (CD) spectroscopy The purified RgCrtC and TrCrtC were diluted to 0.04 and 0.03 mg ml-1, respectively, in 10

mM Na2HPO4, pH 8.0. Samples were incubated for 5 min at temperatures from 5 to 90ºC (5°C steps) and after each incubation samples were scanned. CD spectra were collected from 190 to 250 nm as an average of five spectra, with a data pitch of 1 nm. A band width of 1 nm was used with a detector response time of 0.25 sec and scanning speed of 100 nm min-1. CD spectra were recorded on a Jasco J-810 spectrometer equipped with a Peltier temperature control unit in 0.1 cm path length cuvette [20].

39 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina 2.2.9 Metal analysis using USN-ICP-OES The metal content from purified protein sample and the buffer solution was measured using Perkin-Elmer 4300 dual view inductively coupled plasma (ICP) with optical emission spectroscopy (OES) spectrometer, coupled with ultrasonic nebulizer (USN) U-6000 AT, Cetac. Measurements were performed for different metals and at different wavelengths, as following: Co (228 nm and 238 nm), Fe (238 nm and 239 nm), Mo (202 nm and 203 nm), Ni (231 nm and 221 nm) and Zn (206 nm and 213 nm).

2.3 Results

2.3.1 Expression and purification of the carotenoid 1,2- hydratases For biochemical characterization of the carotenoid 1,2-hydratases and comparison of their catalytic activities, the two crtC genes from R. gelatinosus and T. roseopersicina were

cloned into the expression vector pET15-b. The constructed pET15b_CrtCRg and

pET15b_CrtCTr plasmids were sequenced and the results confirmed that the genes were

successfully inserted in frame with the N-terminal His6-tag. In order to express the recombinant CrtC’s, E. coli BL21 (DE3) was transformed with the expression plasmids. SDS/PAGE analysis revealed in both cases a 44 kDa band (Figure 2.1), which is consistent with the value calculated from the deduced amino acid sequence.

Figure 2.1 SDS-PAGE (10%) analysis of expression and IMAC purification for RgCrtC (lane 1-3) and TrCrtC (lane 4-6). M: precision plus protein standard; a: whole cells before induction; b: whole cells after induction with 0.1 mM IPTG and expression overnight at 25ºC; c: purified CrtC’s.

The expression level of RgCrtC was around 2 times higher than that of TrCrtC expressed under the same conditions. In the case of TrCrtC an additional faint band around 38 kDa was detected after induction (Figure 2.1, lane 5), which is absent in the non-induced sample (Figure 2.1, lane 4).

40 2.3 Results

CrtC’s were purified from CFE’s by a single step IMAC column and led to a nearly homogenous band of 44 kDa in the case of RgCrtC and bands of 38 kDa and 44 kDa in the case of TrCrtC (Figure 2.1, lanes 3 and 6). However, the larger band could not be detected again, once the sample was stored at -20ºC for a few days (Figure 2.1, lane 6).

2.3.2 Hydratase activity Activity measurements of the purified enzymes with lycopene as substrate demonstrated functional expression of the recombinant CrtC’s in E. coli. The purified enzymes catalyze

the conversion of lycopene into both 1-HO-lycopene and 1,1′-(HO)2-lycopene. For both CrtC’s the conversion rate was 30% and the ratio between mono- and dihydroxylated product was 2:1. USN-ICP-OES metal analysis showed that the protein samples did not contain any significant amounts of iron, zinc, , nickel or molybdenum (data not shown). Furthermore, it was observed that the addition of coenzymes NAD+/NADH or protease inhibitors had no detectable influence on enzyme activity. Although the effect of various metal ions on the hydratase activity was tested, no firm conclusion could be drawn from these data as the metals have a degrading effect on the substrate lycopene [21].

2.3.3 Enzyme kinetics In order to compare the catalytic activities of the two expressed CrtC’s, in vitro activity studies were performed. Since the conversion of lycopene to 1-HO-lycopene and 1,1′-

(HO)2-lycopene with isolated enzyme was very slow, the reactions were carried out with

CFE’s (Figure 2.2). Kinetic parameters Km, Vmax, Vmax/Km and kcat/Km were determined by activity assay using lycopene as substrate at 28ºC (Table 2.1). The results are shown in a Michaelis-Menten plot (Figure 2.3) as the reaction rate versus the substrate concentration.

Table 2.1 Kinetic parameters for recombinant Rubrivivax gelatinosus CrtC and Thiocapsa roseopersicina CrtC

-1 -1 2 -1 -1 Name Vmax (nmol h mg ) Km (µM) Vmax/Km (x 10 ) kcat/Km (h nmol ) RgCrtC 0.32 ± 0.08 24.7 ± 12.7 1.3 0.57 TrCrtC 0.15 ± 0.02 9.8 ± 4 1.6 0.71

41 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

(A)

(B) 3 RgCrtC 5000 TrCrtC

4000 1

2 3000

2000 Intensity [mV]

1000

0 0246810 Time [min]

Figure 2.2 Reaction catalyzed by Rubrivivax gelatinosus and Thiocapsa roseopersicina carotenoid 1,2- hydratase; the conversion of lycopene into 1-HO-lycopene and 1,1′-(HO)2-lycopene (A). HPLC separation of carotenoids formed in vitro by E.coli extract expressing the RgCrtC (solid line) and TrCrtC (dashed line).

Peak 1, 1,1′-(HO)2-lycopene; peak 2, 1-HO-lycopene; peak 3, lycopene (B).

The affinity constant (Km) for recombinant RgCrtC and TrCrtC was calculated as 24 and -1 -1 9.5 µM, respectively, and Vmax was 0.31 and 0.15 nmol h mg , respectively. The substrate

specificity values were calculated as Vmax/Km and the results show a slightly higher specificity of TrCrtC with 1.6 x 102 compared to RgCrtC with 1.3 x 102 for lycopene. Furthermore, the catalytic efficiency values for TrCrtC (0.71 h-1 nmol-1) and RgCrtC (0.57 h-1 nmol-1) revealed no significant difference for lycopene hydration.

42 2.3 Results

0,20

0,15 ] -1 mg -1 0,10 V [nmol h V [nmol 0,05

RgCrtC TrCrtC 0,00 0 5 10 15 20 25 30 35 40 Lycopene [M]

Figure 2.3 Michaelis-Menten plot of recombinant RgCrtC (●) and TrCrtC (○). The cell-free extracts were assayed with various lycopene concentrations (0.5-40 µM) in 50 mM Na2HPO4 sodium phosphate (pH 8.0) at 28ºC for 4 h. The rates of product formation (1-HO-lycopene plus 1,1′-(OH)2-lycopene) are plotted against varying substrate concentrations. Kinetic constants are listed in Table 2.1.

2.3.4 Substrate specificity Substrate specificity was tested with acyclic alkenes of different chain length, which possess the same alkenyl functional group like lycopene, the natural substrate of CrtC (Supplementary figure 2.1). No activity was detected for the C5, C6 and C15 substrate using standard assay conditions. However, a product was detected with the C20 substrate geranylgeraniol for both RgCrtC and TrCrtC, which was absent in the control experiment. The conversion was very low, approximately 5% (Supplementary figure 2.2).

2.3.5 Effect of pH and temperature on hydratase activity and stability The dependence of the activity of recombinant RgCrtC and TrCrtC at different pH values and temperatures was investigated using lycopene as substrate. The optimum pH for hydratase activity appeared to be pH 8.0 (Figure 2.4A). While RgCrtC has a broader pH optimum ranging from pH 7.0-8.0, a significant decrease was observed for TrCrtC with

43 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina only 50% activity at pH 7.0. No activity could be detected at pH 4.0-5.0 for both enzymes. At higher pH values, both enzymes showed rapid decrease of activity, although 50% of the relative activity was still detected at pH 9.0. Both enzymes retained much residual activity after 30 min incubation at pH 4.0-8.0, indicating that the CrtC’s are stable in both slightly alkaline and acid environments (Supplementary figure 2.3A). However, compared with RgCrtC, the TrCrtC stability decreases in the range from pH 6.5-9.0 to only 56% of the residual activity, whereas RgCrtC still remains 95% activity at pH 9.0. Despite no detected activity at pH 4.0-5.0 (Figure 2.4A) both enzymes seem to be stable at that pH range and still show 75%-80% of residual activity. The effect of temperature on hydratase activity from 0 to 50ºC is depicted in Figure 2.4B. The favorable temperature range was from 25 to 35ºC with an optimum at 30ºC. Enzyme activity for RgCrtC and TrCrtC was significantly lower at 20ºC (55 and 42%, respectively) and 40ºC (47 and 31%, respectively). A negligible activity was found at 5ºC (around 10%). Thermal stability was investigated by pre-incubating hydratases for 30 min at different temperatures and subsequently testing residual activity under standard assay conditions (Supplementary figure 2.3B).

44 2.3 Results

(A) 0,20 RgCrtC TrCrtC ] -1 0,15 mg -1

0,10

0,05 Enzyme activity [nmol h Enzyme

0,00 456789 pH

(B)

0,20 RgCrtC TrCrtC ] -1 0,15 mg -1

0,10

0,05 Enzyme activity [nmol h

0,00 0 102030405060 Temperature [ºC]

Figure 2.4 Effect of pH (A) and temperature (B) on activity of RgCrtC (●) and TrCrtC (○). For pH effect measurements were performed with lycopene under standard assay conditions using different buffers: 100 mM acetate (pH 3.6, 4.0 and 5.0), 100 mM potassium phosphate (pH 6.0, 7.0 and 8.0) and 50 mM Tris-HCl (pH 8.6 and 9.0). For temperature effect activity assays were performed with lycopene at various temperatures (1-50ºC) under standard assay conditions.

45 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

Maximum stability was recorded at 5ºC. The enzymes did not show significant decrease of the activity up to 40ºC. When pre-incubated at 45ºC they still showed relative high activity (around 50 to 60%). However, RgCrtC was extremely sensitive at 50ºC retaining only 6% activity after 30 min pre-incubation, while TrCrtC showed 30% residual activity at that temperature. Additionally, the temperature stability of RgCrtC and TrCrtC was studied using CD spectroscopy. It was not possible to obtain a good qualitative representation of the CD spectra from RgCrtC, in contrast to TrCrtC, which is shown in Figure 2.5.

0

-5

-10 CD/CD [mdeg]

-15 5ºC 50ºC 70ºC 90ºC -20 200 205 210 215 220 225 230 235 240 245 250 Wavelength [nm]

Figure 2.5 Enzyme stability of recombinant TrCrtC by CD spectroscopy. The purified CrtC was diluted to 0.03 mg/ml with 10 mM sodium phosphate, pH 8.0, and incubated for 5 min from 5 to 90ºC. CD assay was performed by wavelength scan from 190 to 250 nm.

TrCrtC was found to be relatively stable below 50ºC. Significant change in the secondary structure was observed at temperatures above 50ºC, which corresponds with the results obtained in the activity assay (Figure 2.4B).

2.4 Discussion

This study reports on the purification and biochemical characterization of two heterologously expressed carotenoid 1,2-hydratases (CrtC) from photosynthetic bacteria,

46 2.4 Discussion

which are potential biocatalysts in the green hydration of carotenoid-like substrates. The two crtC genes from R. gelatinosus (1221 bp) and T. roseopersicina (1218 bp) were cloned, sequenced, and successfully expressed in E. coli BL21 (DE3). Many attempts have been made to optimize expression levels and to reduce formation of inclusion bodies (data not shown), as these enzymes are detected in the membrane fraction. Hydropathy plots, determined with Kyte-Doolittle [22], did not reveal any putative transmembrane domain in the two hydratases (Supplementary figure 2.5). However, it was noticed that the first 45-55 amino acids of RgCrtC and TrCrtC showed a significantly higher percentage of proline (13 and 16 %, respectively) whereas the rest of the sequence has the usual proline amount of 9 and 8%, respectively. Ouchane and co-workers described this already for RgCrtC [23]. Proline rich regions in proteins are widely found in prokaryotes and [24]. A non-repetitive (XP)n sequence like identified in RgCrtC (10x) and TrCrtC (9x) can have different functions as for instance stabilizing the enzyme by binding noncovalently to other proteins, binding to other hydrophobic structures like hydrophobic substrates or function as a “molecular trigger” passing signals to the inner membrane. Based on these findings one may suggest that the hydrophobic N-terminus of the RgCrtC and TrCrtC could play a role in stabilizing the enzyme in the hydrophobic membrane area. This hypothesis is strengthened by the following data. On the nucleotide level the two crtC sequences presented a relatively high identity of 70 % [25]. However, a significant difference was observed after the heterologous expression in E. coli. Although the gene sequence predicts a protein of 44 kDa (Supplementary figure 2.4B), the SDS/PAGE analysis of the expressed enzymes showed a second band of about 38 kDa for TrCrtC, which was absent in RgCrtC as well as in the empty vector control. Furthermore, membrane fractions with only a visible 38 kDa band showed good activity (data not shown) indicating that the lower band is active. MS data of this band revealed that the N-terminal proline rich part is missing, thereby supporting the hypothesis that this part is not important for biological activity or substrate binding but for membrane association. Moreover, analysis of amino acid sequence similarities of various known and putative CrtC’s also shows that the first part of the sequence is missing in a number of the analyzed sequences (data not shown). As this phenomenon of protein processing is not known from literature to occur in the CrtC family, more experiments were performed. One approach currently under study, that addresses our hypothesis, involves construction of mutants, which lack the N-terminal part of the sequence. First results showed that the truncated CrtC’s are fully functional and catalyze the conversion of lycopene to the corresponding products without any loss of activity

47 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

(Supplementary Figure 2.7). This observation confirms our hypothesis that the N-terminal part is not involved in catalytic reaction nor in substrate binding. Furthermore, this means that the truncated enzyme could be used in an industrial setting.

The RgCrtC and TrCrtC catalyze the conversion of lycopene with a Vmax of 0.31 and 0.15 -1 -1 nmol h mg , respectively, and a Km of 24 and 9.5 µM, respectively, without the need of

any cofactor. The lower value of Km observed for TrCrtC shows that this enzyme presents higher affinity for the substrate lycopene than RgCrtC. However, the catalytic efficiency

values were similar for both enzymes, despite TrCrtC presenting twofold lower Vmax. Maximum activity was detected for both enzymes at pH 8.0 and at the temperature of 30ºC. Moreover, they also presented quite good activities at temperatures ranging from 25 to 35ºC. However, temperatures above 50ºC caused denaturation of the protein structure and therefore inactivation, which was confirmed by CD spectroscopy. Although both enzymes are rather similar in their pH and temperature profile, RgCrtC is more stable at higher pH’s, while TrCrtC is more stable at higher temperatures. This could be of importance when choosing the right enzyme for a biocatalytic process. The substrate scope study of CrtC’s is an important aspect as no investigation has been made in this direction to date. Next to the substrate lycopene, activity measurements were reported in literature with two other natural substrates neurosporene and spheroidene, as demonstrated by Steiger et al. [16]. It was concluded that spheroidene, which possess a terminal methoxy functional group, serves as the best substrate for RgCrtC. Furthermore, this enzyme was also able to use monohydroxy carotenoids as substrates, which could not be observed for Rhodobacter capsulatus CrtC [16]. Our primary objective with the substrate specificity experiment was to investigate the possibility of using CrtC with unnatural substrates to produce highly valuable compounds for industry. Based on the observed activity with geranylgeraniol we postulate that the minimum size of the substrates for both RgCrtC and TrCrtC is C20 (twenty carbon) chain. However, the low conversion of about 5% clearly indicates that their substrate spectrum is limited. Since the crystal structure of CrtC has not yet been solved, one can only speculate about the size of the active site and the mechanism that is involved in the hydration of the substrates. Further structural and biochemical characterization is necessary to achieve a full understanding of this enzyme and its reaction mechanism.

48 2.5 Acknowledgments

In conclusion, both CrtC’s are stable at a broad and suitable temperature and pH range and hydrate several long aliphatic substrates to give tertiary alcohols. Future studies will be directed at improving the activity of these hydratases.

2.5 Acknowledgments

We thank Prof. Dr. Gerhard Sandmann and Prof. Dr. Kornél L. Kovács for providing the plasmids. This project is financially supported by the Netherlands Ministry of Economic Affairs and the B-Basic partner organizations (www.b-basic.nl) through B-Basic, a public- private NWO-ACTS programme (ACTS = Advanced Chemical Technologies for Sustainability).

49 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina 2.6 Supplementary information

Supplementary table 2.1 Bacterial strains and plasmids used in this study Strain and plasmid Relevant trait(s) Source or reference Strains

– - - E. coli BL21 (DE3) F ompT gal dcm lon hsdSB(rB mB ) λ(DE3) Novagen E. coli TOP10 F- mcrA Δ(mrr-hsdRMS-mcrBC) φ80lacZΔM15 Invitrogen ΔlacX74 nupG recA1 araD139 Δ(ara-leu)7697 galE15 galK16 rpsL(StrR) endA1 λ- Plasmids

pPQE30crtCRg pPQ30; carries the BamHI / KpnI fragment with [16] crtC from Rvi. gelatinosus pTcrt3 pBluescript SK(+); carries the wild-type BamHI- [11] SacI fragment of the crtDC operon of Tca. roseopersicina pET15-b E. coli general expression vector with N- Novagen terminal His tag; Ampr

pET15b_CrtCRg pET15-b with 1252-bp NdeI / XhoI fragment this work

from pPQE30crtCRg

pET15b_CrtCTr pET15-b with 1249-bp NdeI / XhoI fragment this work from pTcrt3

50 2.6 Supplementary information

Supplementary figure 2.1 Structure of substrates used for substrate specificity studies of Rvi. gelatinosus and Tca. roseopersicina carotenoid 1,2-hydratase.

(x1,000,000)

4.75

4.50

4.25

4.00

3.75

3.50

3.25

3.00

2.75

2.50

2.25

2.00

1.75

1.50

1.25

1.00

0.75

0.50

0.25

0.00 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0

Supplementary figure 2.2 GC separation of products formed in vitro by Escherichia coli extract expressing RgCrtC (pink line) and TrCrtC (blue line) using the substrate geranylgeraniol. Obtained product is indicated with arrow (RT 41 min). Extract with empty plasmid pET15-b served as negative control (black line).

51 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

(A) 0,04 RgCrtC TrCrtC ]

-1 0,03 mg -1

0,02

0,01 Enzyme activity [nmol h

0,00 345678910 pH

(B) 0,20 RgCrtC TrCrtC ]

-1 0,15 mg -1

0,10

0,05 Enzyme activity Enzyme [nmol h

0,00 0 102030405060 T [ºC]

Supplementary figure 2.3 Stability of RgCrtC (●) and TrCrtC (○) at different pH (A) and temperature (B) values. The remaining activity was assayed under standard assay conditions after the cell-free extracts had been incubated in the corresponding buffers (pH 3.6 to pH 9.0) or at the indicated temperature (5-50ºC) in 50 mM Na2HPO4 sodium phosphate (pH 8.0) for 30 min

52 2.6 Supplementary information

Supplementary figure 2.4 Amino acid sequence alignment of RgCrtC and TrCrtC (Clone Manager 9 Professional Edition). Identical amino acids are highlighted in red.

53 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

(A)

(B)

Supplementary figure 2.5 Hydropathy plot of the RgCrtC (A) and TrCrtC (B) amino acid sequence. Hydropathy scores [22] for a window of 19 residues were averaged and assigned to the first amino acid of the window. A hydropathy score greater than 1.6 indicates transmembrane region.

54 2.6 Supplementary information

Supplementary figure 2.6 DNA sequence alignment of RgCrtC and TrCrtC (Clone Manager 9 Professional Edition). Identical bases are highlighted in red.

55 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

(A) M 1a 1b 2a 2b 3a 3b 4a 4b 5a 5b

50

37

(B)

Supplementary figure 2.7 SDS-PAGE (10%) analysis (A) and amino acid sequence alignment of the wildtype and the truncated CrtC (B). M: precision plus protein standard; a: whole cells before induction; b: whole cells after induction with 0.1 mM IPTG and expression overnight at 25°C; 1: pET15-b; 2: TrCrtC wildtype; 3: TrCrtC truncated; 4: RgCrtC wildtype; 5: RgCrtC truncated.

56 2.7 References

2.7 References

[1] R. Kourist, et al., "Enzymatic synthesis of optically active tertiary alcohols: Expanding the biocatalysis toolbox," Chembiochem, vol. 9, pp. 491-498, Mar 3 2008. [2] P. G. Cozzi, et al., "Enantioselective catalytic formation of quaternary stereogenic centers," European Journal of Organic Chemistry, pp. 5969-5994, Dec 2007. [3] P. Fernandes, "Miniaturization in Biocatalysis," International Journal of Molecular Sciences, vol. 11, pp. 858-879, Mar 2010. [4] D. Brady, et al., "Characterisation of nitrilase and nitrile hydratase biocatalytic systems," Applied Microbiology and Biotechnology, vol. 64, pp. 76-85, Mar 2004. [5] K. Rzeznicka, et al., "Cloning and functional expression of a nitrile hydratase (NHase) from Rhodococcus equi TG328-2 in Escherichia coli, its purification and biochemical characterisation," Applied Microbiology and Biotechnology, vol. 85, pp. 1417-1425, Feb 2010. [6] S. Gocho, et al., "BIOTRANSFORMATION OF OLEIC-ACID TO OPTICALLY-ACTIVE GAMMA-DODECALACTONE," Bioscience Biotechnology and Biochemistry, vol. 59, pp. 1571- 1572, Aug 1995. [7] A. Wanikawa, et al., "Detection of gamma-lactones in malt whisky," Journal of the Institute of Brewing, vol. 106, pp. 39-43, Jan-Feb 2000. [8] L. E. Bevers, et al., "Oleate hydratase catalyzes the hydration of a nonactivated carbon-carbon bond," Journal of Bacteriology, vol. 191, pp. 5010-5012, Aug 2009. [9] G. A. Armstrong, et al., "Nucleotide-sequence, organization, and nature of the protein products of the carotenoid biosynthesis gene-cluster of Rhodobacter-capsulatus," Molecular & General Genetics, vol. 216, pp. 254-268, Apr 1989. [10] H. P. Lang, et al., "Complete DNA-sequence, specific Tn5 insertion map, and gene assignment of the carotenoid biosynthesis pathway of Rhodobacter-sphaeroides," Journal of Bacteriology, vol. 177, pp. 2064-2073, Apr 1995. [11] A. T. Kovacs, et al., "Genes involved in the biosynthesis of photosynthetic pigments in the purple sulfur photosynthetic bacterium Thiocapsa roseopersicina," Applied and Environmental Microbiology, vol. 69, pp. 3093-3102, Jun 2003. [12] E. Giraud, et al., "Two distinct crt gene clusters for two different functional classes of carotenoid in Bradyrhizobium," Journal of Biological Chemistry, vol. 279, pp. 15076-15083, Apr 9 2004. [13] N. U. Frigaard, et al., "Genetic manipulation of carotenoid biosynthesis in the green sulfur bacterium Chlorobium tepidum," Journal of Bacteriology, vol. 186, pp. 5210-5220, Aug 2004. [14] J. A. Botella, et al., "A cluster of structural and regulatory genes for light-iduced carotenogenesis in Myxococcus-xanthus," European Journal of Biochemistry, vol. 233, pp. 238-248, Oct 1995. [15] Z. T. Sun, et al., "A novel carotenoid 1,2-hydratase (CruF) from two species of the non- photosynthetic bacterium Deinococcus," Microbiology-Sgm, vol. 155, pp. 2775-2783, Aug 2009.

57 Biochemical Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

[16] S. Steiger, et al., "Heterologous expression, purification, and enzymatic characterization of the acyclic carotenoid 1,2-hydratase from Rubrivivax gelatinosus," Archives of Biochemistry and Biophysics, vol. 414, pp. 51-58, Jun 1 2003. [17] N. Igarashi, et al., "Horizontal transfer of the photosynthesis gene cluster and operon rearrangement in purple bacteria," Journal of Molecular Evolution, vol. 52, pp. 333-341, Apr 2001. [18] A. M. Sevcenco, et al., "Development of a generic approach to native metalloproteomics: application to the quantitative identification of soluble copper proteins in Escherichia coli," Journal of Biological Inorganic Chemistry, vol. 14, pp. 631-640, May 2009. [19] S. Steiger, et al., "Substrate specificity of the expressed carotenoid 3,4-desaturase from Rubrivivax gelatinosus reveals the detailed reaction sequence to spheroidene and spirilloxanthin," Biochemical Journal, vol. 349, pp. 635-640, Jul 15 2000. [20] Y. H. Chen, et al., "Determination of secondary structures of proteins by Circular-Dichroism and Optical Rotatory Dispersion," Biochemistry, vol. 11, pp. 4120-&, 1972. [21] C. S. Boon, et al., "Role of Iron and Hydroperoxides in the Degradation of Lycopene in Oil-in-Water Emulsions," Journal of Agricultural and Food Chemistry, vol. 57, pp. 2993-2998, Apr 2009. [22] J. Kyte and R. F. Doolittle, "A Simple Method for Displaying the Hydropathic Character of a Protein," Journal of Molecular Biology, vol. 157, pp. 105-132, 1982. [23] S. Ouchane, et al., "Pleiotropic effects of puf interposon mutagenesis on carotenoid biosynthesis in Rubrivivax gelatinosus - A new gene organization in purple bacteria," Journal of Biological Chemistry, vol. 272, pp. 1670-1676, Jan 17 1997. [24] M. P. Williamson, "The Structure and Function of Proline-Rich Regions in Proteins," Biochemical Journal, vol. 297, pp. 249-260, Jan 15 1994. [25] S. F. Altschul, et al., "Basic Local Alignment Search Tool," Journal of Molecular Biology, vol. 215, pp. 403-410, Oct 5 1990.

58

Chapter 3

3 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

Aida Hiseni, Linda G. Otten and Isabel W.C.E. Arends

Submitted

Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina Abstract

Carotenoid 1,2-hydratases (CrtC) catalyze the selective addition of water to an isolated carbon-carbon double bond. Although their involvement in the carotenoid biosynthetic pathway is well understood, little is known about the mechanism by which these hydratases transform carotenoids such as lycopene into corresponding hydroxyl compounds. To gain insight into the enzymatic mechanism of CrtC’s point mutants of selected conserved amino acids were generated. Rubrivivax gelatinosus CrtC point mutants in which each of the amino acids His239, Trp241, Tyr266 and Asp268 were individually changed into Ala, and the corresponding point mutants of Thiocapsa roseopersicina CrtC, were completely inactive. This result suggests the identification of key residues which are directly involved in the catalytic reaction. Furthermore, the analysis of a partial 3D structure of CrtC, which was obtained by homology modeling with the putative AttH protein from Nitrosomonas europaea, supported these results as all amino acids were in close distance to each other. These results for the first time shed light on the potential catalytic mechanism of CrtC.

60 3.1 Introduction

3.1 Introduction

Carotenoids, which represent one of the most abundant natural pigments with structural and protective properties [1], play an essential role in the photosynthetic machinery of phototrophic organisms such as purple bacteria [2] and higher plants [3]. In addition, they have been identified in fungi and some non-photosynthetic bacteria [4]. Carotenoid 1,2- hydratase (also known as CrtC) is a member of hydro-lyases group EC 4.2.1.131. The enzyme takes part in the biosynthetic pathway of carotenoids [5]. CrtC introduces a tertiary hydroxyl group into an acyclic carotenoid molecule by addition of water to the carbon- carbon double bond at the C-1 position. The enzyme belongs to the Pfam family PF07143 that encompasses members from several purple photosynthetic bacteria. On the other hand, CrtC’s have been identified, which are able to hydrate mono-cyclic carotenoid gamma- carotene. These are evolutionary very distinct from the PF07143 members [6] and they have been given the name CruF. Recently, two representatives of the PF07143 family, the CrtC’s from purple non-sulfur Betaproteobacteria Rubrivivax gelatinosus and purple sulfur Gammaproteobacteria Thiocapsa roseopersicina, respectively, were recombinantly expressed and characterized [7]. Biochemical studies have revealed that these enzymes are able to convert cofactor independently lycopene into 1-HO-lycopene and 1,1’-(HO)2-lycopene (Figure 3.1).

Figure 3.1 Reaction catalyzed by Rubrivivax gelatinosus and Thiocapsa roseopersicina carotenoid 1,2- hydratase; the conversion of lycopene into 1-HO-lycopene and 1,1′-(HO)2-lycopene.

In addition, they showed some activity towards the unnatural substrate geranylgeraniol, a C20 molecule that resembles the natural substrate lycopene (Figure 3.2).

Figure 3.2 Molecular structure of geranylgeraniol.

61 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

CrtC’s are appealing enzymes in the biotechnology field because they are able to make a tertiary alcohol, a highly valuable building block for the synthesis of several bioactive natural products and pharmaceuticals [8]. Furthermore, they possess an intrinsically high stability at a wide pH and temperature range, which constitute useful properties for an industrial application [7]. The subcellular location of this enzyme in the cell membrane fraction (membrane-bound) allows for a straightforward isolation and simplified large scale purification. From a chemical point of view, CrtC’s are able to perform a challenging chemical reaction, namely the selective addition of water to an isolated carbon-carbon double bond [9]. The reaction proceeds without assistance of electron withdrawal groups, or transition metal cations and in vitro requires harsh acidic conditions [10]. Furthermore, the CrtC’s from photosynthetic bacteria act on acyclic carotenoids, whereas the CruF’s from non- photosynthetic bacteria catalyze the hydration of mono-cyclic carotenoids. Nevertheless, the catalytic and structural features, that determine hydratase activity and specificity of these two distinct families remains unknown. To our knowledge no published data exist on the catalytic mechanism of this group of enzymes, nor has the 3D structure been elucidated yet. However, the 3D structure of the first representative of the Pfam family PF09410 (putative AttH) has been solved, a family which is distantly related to the CrtC family PF07143 [11]. The mechanism of lycopene hydration, which involves proton attack at C-2 and C-2′ with a carbonium ion intermediate 2 and the introduction of the hydroxyl group at C-1 and C-1′, was established from H2O- labeling studies with intact cells [12, 13]. For a hydration reaction, it is likely to assume that the first step in the reaction is protonation of the alkene, leading to an intermediate carbocation. Quenching of the carbocation by water will lead to the alcohol as product. The protonation of hydrophobic long-chain alkenes has also been described for the enzyme class of cyclases, of which the full mechanism is known [14, 15]. The objective of this study was to provide an insight into the possible hydration mechanism of CrtC’s. Based on the better knowledge of the mechanistic reaction, it might be possible to improve enzyme activity or substrate scope by for instance directed evolution or (semi)rational design. Through multi-sequence alignment of several CrtC homologues, highly conserved amino acids were identified, which could be functionally or structurally important. The corresponding alanine mutants of these amino acids were produced and in this way their involvement in the hydratase activity could be evaluated. Furthermore, a

62 3.2 Materials and methods

homology model of CrtC was obtained by using the putative AttH protein from Nitrosomonas europaea as a template [11]. Following the identification of catalytically active amino acid residues the aim was to propose a catalytic mechanism for CrtC catalyzed water addition.

3.2 Materials and methods

3.2.1 In silico analysis BLAST [16] was used to select carotenoid 1,2-hydratase homologues. In order to look for identities/similarities between the CrtC homologues, nucleotide and amino acid sequences were aligned with the BioEdit Sequence Alignment Editor v.7.1.3.0 (www.mbio.ncsu.edu/bioedit/bioedit.html). In addition, protein sequences were subjected to protein functional analysis using Conserved Domain Search (CDD) [17] and Pfam search [18], and a protein phylogenetic tree was constructed with Phylogeny.fr [19, 20]. The CrtC secondary structure prediction was carried out with the program PolyView 2D [21]. The SWISS-MODEL program was used to model the structure of CrtC [22].

3.2.2 Cloning of carotenoid 1,2-hydratase genes

Plasmids pET15b_CrtCRg and pET15b_CrtCTr containing CrtC from R. gelatinosus (Rg) and T. roseopersicina (Tr), respectively, were constructed in a previous study [7]. Two fosmids with crtC genes from metagenomic samples DelRiverFos06H03 (Fos06) and DelRiverFos13D03 (Fos13), respectively, were kindly provided by Dr. Kirchman [23]. The cosmid encoding CrtC from Bradyrhizobium (Br) was received from Dr. Dreyfus [24]. In order to get sufficient DNA material for further studies, the fosmid- and cosmid DNA were amplified in E. coli TOP 10 cells. After DNA isolation with the QIAprep Spin Miniprep Kit (Qiagen) from the cells sufficient DNA was obtained for further research. The crtC’s from Rhodospirillum rubrum (Rr) and Rhodopseudomonas palustris (Rp) were amplified from genomic DNA. For that, genomic DNA of R. rubrum (Rr) was kindly provided by Prof. Roberts (NCBI Reference Sequence: NC_007641.1). R. palustris cells (DSM No. 123) were obtained from DSMZ (Deutsche Sammlung von Mikroorganismen und Zellkulturen), enriched in appropriate medium according to DSMZ instructions and gDNA isolated using the UltraClean Soil DNA Isolation Kit (Mobio). Subsequently, primers were designed for the isolation of all crtC genes (Table 3.1), which carry two restriction sites for

63 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

subsequent cloning: NdeI (forward) and XhoI (reverse). For BrcrtC the XhoI-site was replaced with BamHI, because the XhoI-site was present in the gene itself. Amplification reactions were done using standard PCR reactions. Using appropriate restriction sites, the digested and purified fragment was ligated into the same sites of the pET15-b vector and transformed into E. coli TOP10 competent cells. The insertion of the crtC gene was verified by restriction analysis with the corresponding restriction enzymes (New England Biolabs) and DNA sequencing (BaseClear). Table 3.1 Primers used in this study. The respective restriction sites are underlined. Name Sequence (5'→3') Restriction site DRF06 FW GGGAGTACCATATGAGTGATGATGGCCAAC NdeI DRF06 RV ATCCGCTCGAGATAATCTCAAGCCCGCCTCG XhoI DRF13 FW GGGAGTACATATGGATGGCGTGTCAGAC NdeI DRF13 RV CCGCTCGAGTAATGCTTAGGGCCACTTGGC XhoI Br FW CGGACATCATATGTGCCCGCCAG NdeI Br RV ATCCAGGATCCATCGCGTGAACTTCACCACC BamHI Rp FW CGGGACTTCCATATGTCAGGAGCTGAGTTG NdeI Rp RV ACCGCTCGAGTAACGTTCAGCGGAACGC XhoI Rr FW GGGAAATTCCATATGCACCGCCCGGAC NdeI Rr RV GCTCGAGTTCAATTAGCCCTTAACCGCCGC XhoI

3.2.3 Construction of CrtC mutants

3.2.3.1 Single point mutations Single amino acid exchange within the crtC genes of Rg and Tr was done by the megaprimer PCR method introduced by Kammann et al. [25] and later modified by Sarkar and Sommer [26] and Landt et al. [27]. The mismatch primers are listed in Table 3.2. In the first PCR reaction, performed under standard reaction conditions, the megaprimer was produced using the corresponding forward primer containing the desired base substitution (Table 3.2) in combination with the reverse primer Rg_rv [7] and Tr_rv [7], respectively.

Plasmids pET15b_CrtCRg and pET15b_CrtCTr [7] were used as template. The size and purity of the megaprimer was verified by agarose gel electrophoresis. In order to produce the full length gene, a second PCR reaction was performed with the corresponding megaprimer and Rg_fw [7] or Tr_fw [7], respectively. Subsequent steps were performed as described in previous section. The insertion of the crtC gene and the presence of the

64 3.2 Materials and methods desired mutation were verified by restriction analysis with NdeI/XhoI enzymes and DNA sequencing (BaseClear). Table 3.2 Primers for site directed mutagenesis. Mismatch points are underlined. Amino acid Sequence (5'→3') exchange H239A AGCGGCGGACGCGCTCGCTG W241A CATCGCGCGGGGCCGATCG H264A CTGGAGCGGCGCCGCCTACC Y266A GCCACGCCGCCCTCGACT

R. gelatinosus gelatinosus R. D268A CGCCTACCTCGCCTCGAACGAAG H237A GATCCGGCGGAACGCGCAGTCTGGTGG W239A CGCCATGTCGCGTGGCCGATC H262A GCTGGAGCGGCGCTGGCTAT D266A CATGGCTATCTCGCCTCAAA S58V GCGTCCGTCGTCGCGCAGCA

T. roseopersicina S58Q GCGTCCCAGGTCGCGCAGCA

3.2.3.2 N-terminally truncated Rg- and TrCrtC’s Rg- and TrCrtC lacking the first 45 and 57 amino acids, respectively, were constructed using primers Rg_45aa (Table 3.3)/Rg_rv [7] and Tr_57 (Table 3.3)/Tr_rv [7] under standard PCR conditions. Plasmids pET15b_CrtCRg and pET15b_CrtCTr [7], respectively, were used as template. Subsequent steps were performed as described in the section “Cloning of carotenoid 1,2-hydratase genes”. The insertion of the crtC gene was verified by restriction analysis with the corresponding restriction enzymes (New England Biolabs) and DNA sequencing (BaseClear). Table 3.3 Primers used for construction of truncated CrtC’s. The NdeI restriction sites are underlined. Name Sequence (5'→3') Rg_45aa AGTACCATATGGGCGACGCACGGCTGG Tr_57aa AGTACCATATGTCCGTCGCGCAGCAAGG

3.2.4 Recombinant expression of CrtC’s E.coli BL21 (DE3) was the host for the pET15_CrtC plasmids. Cultures were grown at 37°C in Luria–Bertani broth with 100 μg ml−1 ampicillin until an OD600 value of 0.6–0.8 was reached. Unless otherwise stated, protein expression was induced by addition of isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 0.1 mM, followed

65 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

by cultivation at 25°C overnight. The cells were harvested by centrifugation at 10.000 rpm

for 10 min at 4°C (Sorvall), washed once with 50 mM Na2HPO4 buffer (pH 8.0), and suspended in the same buffer. In case of subsequent purification, 10 mM imidazole was added to the buffer. Crude extract (CE) from cultures >100 ml was prepared by adding 1 mg ml−1 lysozyme and incubating the cells for 1 h at 4°C, followed by cell disruption at the pressure of 1.5 kBar (Constant systems, IUL instruments). For cultures <100 ml, the cells were disrupted by sonication for 2 min while immersed in an ice-water bath using the microtip probe of a sonicator (Branson Sonicator Cell Disruptor) set at 50% maximal energy. In an effort to reduce the liquid viscosity caused by DNA molecules, 0.1 mg ml-1 of DNAse was added. With the subsequent centrifugation at 10.000 rpm for 20 min at 4°C, cell-free extract (CFE) and pellet were separated. Protein content of the crude extract was determined by BCA assay (Pierce) with bovine serum albumin as the reference protein.

3.2.5 CrtC purification Rg- and TrCrtC ‘active site’ point mutants were purified from the membrane fraction, while the Tr ‘processing’ mutants (S58V, S58Q) were purified from the CFE. The membrane fraction was obtained after the centrifugation of the CFE for 4 h at 13.200 rpm and 8°C. Prior to the addition of Ni-NTA HisTrap HP (GE Healthcare) (previously equilibrated in

50 mM Na2HPO4 buffer, pH 8.0, with 300 mM NaCl and 10 mM imidazole), to the CFE or membrane sample, the membranes were homogenized by ca. 20 passages through a 25G needle. The mixtures were incubated for 1 h at RT, loaded into a polypropylene tube with

porous disc (GE Healthcare) and washed 3 times with washing buffer (50 mM Na2HPO4 buffer, pH 8.0, with 300 mM NaCl and 75 mM imidazole). Then, the CrtC protein was eluted from the column with elution buffer containing 1 M imidazole (50 mM Na2HPO4 buffer, pH 8.0, with 300 mM NaCl). Enzyme fractions were separated by SDS-PAGE (10% Bis-Tris, BioRad) and visualized by staining with SimplyBlue SafeStain (Invitrogen).

3.2.6 Determination of enzyme activity Enzymatic activities were determined with CE on lycopene and geranylgeraniol (GGOH) according to the method described earlier [7], with few modifications. The assay was carried out with 50 μl CE and 20 μM substrate, and 10 mg ml−1 L-α-phosphatidylcholine in the case of lycopene, in a reaction volume of 200 μl. In addition to the GC-MS method, the GGOH reaction products were also analyzed with a newly developed HPLC method. Prior to the analysis, acetonitrile was added to the reaction mixture in a ratio of 60:40

66 3.3 Results and discussion

(ACN:H2O), the mixtures were shaken vigorously for 1 min and solids removed by centrifugation for 1 min at 13.200 rpm. Separation of the reaction products was performed with a Merck 4.6× 50 mm Chromolith TM SpeedROD RP-18e 2 μm column with

ACN/H2O (60:40, v/v) as the eluent. GGOH and the corresponding products were detected at 214 nm (SPD-20A, Shimadzu).

3.3 Results and discussion

3.3.1 Comparative in silico analysis of crtC genes The RgcrtC nucleotide sequence was subjected to a BLAST search in order to identify sequence similarity in different databases. 184 hits were identified, of which 111 were representatives of Proteobacteria. Although, R. gelatinosus belongs to the Betaproteobacteria, more than 77% of the identified 111 hits were from Alphaproteobacteria and only 11% from Betaproteobacteria. Similarly, Igarashi et al. [28] observed that most of the photosynthesis gene products from R. gelatinosus showed high sequence identities to the gene products of R. palustris, an Alphaproteobacteria member. They explain this occurrence as a result of horizontal transfer of the photosynthesis gene clusters from an ancestral species belonging to the Alphaproteobacteria to that of the Betaproteobacteria. The selection of RgCrtC homologues for this study was based on sequence identity and availability of the corresponding gene construct. They originate from all three Proteobacteria subclasses (Alpha, Beta, Gamma) with two additional constructs originating from metagenomic samples from the Delaware River (USA). Figure 3.1 displays a phylogenetic analysis constructed with protein sequences of the selected CrtC homologues. TrCrtC shows the closest relationship to RgCrtC, followed by BrCrtC (55% and 47% sequence identity, respectively). The combined results of Pfam- and Conserved Domain Search showed that all 7 CrtC’s belong to the PF07143 family consisting of several purple photosynthetic bacterial hydroxyneurosporene synthase (CrtC) proteins.

67 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

Figure 3.1 Rooted phylogenetic tree showing the evolutionary relationship between the selected carotenoid 1,2-hydratases. TrCrtC, Thiocapsa roseopersicina (GI 31621263), BrCrtC, Bradyrhizobium sp. BTAi1 (GI 146403799), F06CrtC, uncultured Proteobacterium DelRiverFos06H03 (GI 61653228), F13CrtC, uncultured Proteobacterium DelRiverFos13D03 (GI 61653190), RpCrtC, Rhodopseudomonas palustris (GI 115515977), RrCrtC, Rhodospirillum rubrum (GI 83574254) and RgCrtC, Rubrivivax gelatinosus (GI 29893477).

CrtC sequences were aligned in order to investigate if there are any conserved group- clusters present (Figure 3.2). Indeed, they showed highly conserved regions of in total 64 amino acids. The conserved amino acids are distributed along the sequence ranging from amino acid residues ~115 to ~405. Interestingly, the N-terminal part of the sequence does not contain any conserved amino acids, indicating that this region is probably not necessary for CrtC activity. This could also be the explanation for the absence and the shorter DNA sequences for Fos06, Fos13 and partly RpCrtC, when compared to RgCrtC or TrCrtC. In addition, we found little amino acid variety in 50 positions, as indicated within the boxes in Figure 3.2. Residues involved in the catalysis tend to be highly conserved in a set of homologous proteins that exhibit the same reaction. On the other hand, sequence insertion and sections of low sequence similarity tend to occur in the less important loop regions [29]. The recognition of conserved blocks in CrtC homologues led to the obvious hypothesis that these regions contain the amino acid residues most important for the hydratase activity, specifically those involved in catalysis and substrate binding.

68 3.3 Results and discussion

Figure 3.2 Multiple sequence alignment showing conserved amino acids of the CrtC protein sequences from various bacteria. Identical amino acids are highlighted in black. Positions with only two different amino acids are surrounded by boxes.

The 3D structure of TrCrtC was built by homology modeling based on the only known 3D structure, which showed some sequence identity (17%) to the CrtC (Figure 3.3). The homologue is the putative AttH protein from Nitrosomonas europaea [11]. It belongs to a of unknown function (DUF2006), which has remote similarity to the family PF07143 encompassing carotenoid 1,2-hydratases. The topology of the CrtC structure shows similarity to lipocalins, proteins that bind and transport small hydrophobic molecules [30]. Lipocalin fold is typically formed by a large, twisted beta-sheet that closes in the back to form a central, internal, ligand-binding cavity. This folding motif is frequently found in porins, transmembrane proteins or in general in proteins that bind hydrophobic ligands/substrates [31]. Depending on the protein and the corresponding function, the bound ligand will be entirely within the cavity or part of the ligand will protrude from the cavity at the surface of the protein.

69 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

Figure 3.3 A homology model of CrtC from Thiocapsa roseopersicina based on the crystal structure of a putative AttH (PDB id: 2ICH) from Nitrosomonas europaea. The ribbon diagrams depict front (left-hand side) and back (right-hand side) view of CrtC. The structure is color-coded from the N-terminus (blue) to the C-terminus (red).

In order to investigate if there is any relationship between the conserved regions and specific locations in the homology model of CrtC, the conserved residues were visualized (Figure 3.4). Six striking sequence motifs were selected (I – VI) and the results show that, indeed, most of the conserved residues are located either at the bottom or upper part of the cavity, with one motif (III) located outside the cavity. This order is very interesting and might indicate regions in the CrtC structure, which are involved in the binding of the substrates, while the other might contain amino acids that are directly involved in the catalytic reaction. From a catalytic point of view, amino acids located in regions IV and V, i.e. aspartic acid (D), tyrosine (Y) and histidine (H), are most probably involved in the catalytic hydration, as these amino acids are commonly involved as active residues in acid- base type catalyzed reactions in the active sites of enzymes (Figure 3.5) [32]. Furthermore, they are all in close distance to each other, which is important for the contact with the substrate. It would thus be interesting to analyze these positions by mutagenesis, in order to confirm their involvement in the catalytic process of CrtC’s.

70 3.3 Results and discussion

I

VI

V

III

C

II

IV

N I II III IV V IV VI RgCrtC …...... SDDG…….GSVFSP.Y……...... GPS...... H.W...... H.Y.D...... PFY TrCrtC …...... SDDG…….GSVFSP.Y……...... GPS...... H.W...... H.Y.D...... PFY BrCrtC …...... SDDG…….GSVFSP.Y……...... GPS...... H.W...... D.Y.D...... PFY RrCrtC …...... SDDG…….GSVFSP.Y……...... GPS...... H.W...... H.Y.D...... PFY RpCrtC …...... SDDG…….GSVFSP.Y……...... GPS...... H.W...... H.Y.D...... PFY Fos13CrtC...... SDDG…….GSVFSP.Y……...... GPS...... H.W...... H.Y.D...... PFY Fos06CrtC...... SDDG…….GSVFSP.Y……...... GPS...... H.W...... H.Y.D...... PFY

Figure 3.4 CrtC conserved residues. Ribbon diagram of TrCrtC with marked regions that contain highly conserved amino acid residues. The sequence motifs, which correspond to those regions, are shown in boxes (I – VI). N and C indicate N-and C-terminus, respectively.

71 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

Figure 3.5 View of the potential active site of TrCrtC. The conserved residues H237 (yellow), W239 (red), Y264 (blue) and D266 (green) are show as sticks, with hydrogen bond as yellow dots (region IV).

3.3.2 Production of recombinant wildtype and mutant CrtC’s and enzymatic activity 6 out of the 7 selected CrtC’s were overexpressed from pET15b in E. coli (Figure 3.6). Bands with apparent molecular weight of 32 kDa (Fos13CrtC), 38 kDa (RpCrtC) and 44 kDa (Rr-, Br-, Tr- and RgCrtC) were visualized on SDS-PAGE and were consistent with the values calculated from the deduced amino acid sequences. TrCrtC is expressed as 44- and 38 kDa protein [7].

M 1a 1b 2a 2b 3a 3b 4a 4b 5a 5b 6a 6b 7a 7b 8a 8b M

50

37

25

Figure 3.6 SDS-PAGE (10%) analysis of CrtC expression in E. coli BL21. M, Precision plus protein standard. First lane (a) of each sample shows cells before induction with 0.6 mM IPTG and the second lane (b) shows cells after 4 h expression at 37°C. 1, pET15-b control; 2, Fos06CrtC (32 kDa); 3, Fos13CrtC (32 kDa); 4, RpCrtC (38 kDa); 5, RrCrtC (44 kDa); 6, BrCrtC (44 kDa); 7, TrCrtC (44 kDa); 8, RgCrtC (44 kDa). The indicated molecular weights are deduced from amino acid sequences. CrtC expression bands are indicated by arrows.

72 3.3 Results and discussion

No expression band could be identified for Fos06CrtC. Although, relatively good expression was achieved for most of the CrtC’s, only two were active with lycopene as substrate (data not shown). The fact that all CrtC’s share highly conserved regions in the amino acid sequence (Figure 3.2) indicates that they are performing the same or similar biochemistry. However, no activity whatsoever could be detected for 5 CrtC’s in the standard lycopene hydration assay. At this point it is unclear whether this is due to reasons of low activity in the cell-extract and/or substrate specificity. In our previous study, the two active CrtC’s from R. gelatinosus and T. roseopersicina were biochemically characterized and we showed that both CrtC’s have the ability to convert acyclic carotenoid lycopene into hydroxyl derivatives [7]. Furthermore, we reported on the activity of both CrtC’s with the substrate geranylgeraniol, a C20 acyclic alkene molecule containing a hydroxyl group at one end and -group (acyclic C9 end group according to nomenclature of carotenoids) at the other end of the chain [7]. Unfortunately, the product could not be identified due to low yields. In order to get more insight into the hydration mechanism of CrtC’s, selected amino acid residues in regions IV and V (Figure 3.4) were substituted by the amino acid alanine. The selection was based on the fact that amino acids such as aspartic acid and histidine occur more frequently in enzyme active sites than others [32]. In addition, truncated (Tr- and RgCrtC) and N-terminal point mutants (TrCrtC) were constructed and analyzed. In our previous study we have shortly discussed the preliminary results on the importance of the N-terminal part of CrtC for the catalytic activity. The activity of the truncated versions was fully retained, thus indicating that this part of the enzyme is not essential for activity [7]. Furthermore, the observed cleavage of TrCrtC also supports this hypothesis. Despite the still unknown reason for this occurrence, we were able to identify the cleavage site between S57 and S58 by using MS analysis [7]. In order to exclude any protease background activity from the expression host E. coli, the S58 position was modified by substitution with valine (same size, different chemical features) and glutamine (different size, same chemical features). All mutants (Table 3.2, 3.3) were successfully cloned and expressed in E. coli BL21 (Figure 3.7A). However, clear difference in expression levels was observed. Therefore, all mutants were purified from the membrane fraction in order to ensure that CrtC was present in the cells. As can be seen in Figure 3.7B, all mutants could be purified and showed a band at 38- or 44 kDa, which was absent in the control sample (pET15b). The introduction of mutations and modification of the protein lengths clearly has an effect

73 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina on the expression. While the removal of the N-terminus resulted in an increased expression level, all point mutations negatively influenced the expression of the protein. In the case of the wild type TrCrtC, the purification usually has to be performed as soon as it is expressed, preferably before the cleavage of the N-terminal part (including the His-Tag). This was not the case here and therefore only a very weak protein band is detected after the purification. It should be noted here that the truncated version of TrCrtC shows a larger molecular weight (Figure 3.7, lane 9) compared to the ‘cleaved’ versions (Figure 3.7, lanes 10-15), which is due to the chosen primer position and the attached His-Tag.

M 1 2 3 4 5 6 7 C M

50

37

M 8 9 10 11 12 13 14 15 M

50

37

(A) M 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 C M

(B)

Figure 3.7 SDS-PAGE (10%) analysis of expression (A) and IMAC purification from membrane (B) of CrtC’s from R. gelatinosus (RgCrtC) (lane 1-7) and T. roseopersicina (TrCrtC) (lane 8-15) wildtype and mutants. M, Precision plus protein standard. C, pET15b control. (A) First lane of each sample shows cells before induction with 0.1 mM IPTG and the second lane shows cells after overnight expression at 25°C. 1, RgCrtC wildtype; 2, RgCrtC truncated; 3, RgCrtC H239A; 4, RgCrtC W241A; 5, RgCrtC H264A; 6, RgCrtC Y266A; 7, RgCrtC D268A; 8, TrCrtC wildtype; 9, TrCrtC truncated; 10, TrCrtC S58V; 11, TrCrtC S58Q; 12, TrCrtC H237A; 13, TrCrtC W239A; 14, TrCrtC H262A; 15, TrCrtC D266A.

With regard to the N-terminal point mutations, it seems that the cleavage rate increased in the order of wildtype < S58V < S58Q. This conclusion is based on the fact that in the

74 3.3 Results and discussion

wildtype sample in Figure 3.7A mainly the 44 kDa band is visible (lane 8). In the S58V mutant about the same amount of both, the 38- as well as the 44 kDa bands could be detected (lane 10), while in the S58Q mutant mainly the 38kDa band could be identified (lane 11). Nevertheless, additional experiments are needed to get more insight into this phenomenon. For example, by expressing larger amounts of the corresponding proteins, and by purifying them, one could follow the change of the protein size in time. With the wildtype TrCrtC, we have already shown that even though the purified enzyme consisted of both sizes proteins, after few days of storage, the 44 kDa could not be detected anymore (data not shown). Next to the analysis of the expression levels, the activities of all constructed mutants were measured with lycopene as substrate (Figure 3.8).

4 (A) 3 ]

-1 2

1

0 wt trunc H239A W241A H264A Y266A D268A 4 (B) 3

2 Enzyme activity [nmol mg [nmol Enzyme activity

1

0 wt trunc S58V S58Q H237A W239A H262A D266A

Figure 3.8 Enzymatic activity of wildtype (wt) and mutant CrtC from R. gelatinosus (A) and T. roseopersicina (B). Extracts from E.coli cells expressing the respective enzymes were assayed with 20 μM lycopene in 50 mM Na2HPO4 sodium phosphate (pH 8.0) at 28ºC overnight. Trunc, variants with missing N- terminal residues 1-45 (RgCrtC) and 1-57 (TrCrtC).

As the expression levels were very low for some of the mutants and the activity of CrtC in general is very low, crude extracts were used for the activity assay. Consequently, the results cannot be quantitatively compared. However, in combination with the expression

75 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

levels as shown in Figure 3.7A, indicative conclusions can be drawn. As stated in our previous study the N-terminus is not important for catalytic activity [7]. This again was proven here, as the truncated Rg- as well as TrCrtC were active. Furthermore, the introduction of N-terminal point mutations did not affect the activity of TrCrtC, although it did affect the cleavage rate of the N-terminal part. This might be explained by the fact that the substitution of an amino acid by a smaller or chemically different amino acid could result in conformational changes which promote or prevent the processing activity, either by host proteases or through self-cleavage. On the other hand, it appears that four key residues were identified, which have a potentially important role in the hydration mechanism. By replacing each of the amino acids H239, W241, Y266 and D268 individually by an alanine in RgCrtC the activity is completely destroyed. The same mutations of the corresponding amino acids in TrCrtC, i.e. H237, W239 and D266, also resulted in CrtC inactivation. Unfortunately, the mutagenesis of Y264 in TrCrtC was not successful, and therefore, could not be included in this study. However, based on all the results, one could expect that the mutation of Y264 in TrCrtC would lead to inactivation, as has been seen for RgCrtC. On the other hand, the less conserved H264 in RgCrtC and the corresponding histidine residue in TrCrtC (H262), seem not to have any functional role. The mutants fully retained activity, and even showed slightly increased activity when the expression levels were considered. For instance, the truncated TrCrtC and H262A mutant showed almost the same level of expression (Figure 3.7A, lanes 9 and 14) but the activity of H262A mutant was ~ 1.3-fold higher (Figure 3.8B). The same was observed for RgCrtC, where the expression of the wildtype is much more than that of the mutant H264A, but both showed approximately the same activity. All created mutants were also tested for the activity towards geranylgeraniol. In Figure 3.9, representative HPLC results are depicted. We could confirm activity for mutants that were still catalytically active for lycopene. The formation of the product was followed for two days and an increase was observed, which indicates the activity of CrtC (Table 3.4). In addition, when twice the amount of CrtC was used, the relative amount of the product also increased 2-fold (data not shown). As described in chapter 2, it was not possible to isolate the obtained product in amounts which are necessary for further analysis. However, from the HPLC results we can conclude that the formed product is more hydrophilic than the substrate itself, as it eluted earlier from the reversed phase C-18 column. If our assumption, that CrtC recognizes only a specific part, i.e. -end group, in the carotenoid substrate

76 3.3 Results and discussion

molecule is correct, then one can imagine that CrtC would be able to hydrate the terminal double bond of geranylgeraniol.

600000 blank 2 pET15b 500000 RgCrtC

400000

300000

Intensity [mV] 200000

100000 1

012345678910 Time [min]

Figure 3.9 HPLC separation of diterpene alcohols formed in vitro by E.coli extract expressing the RgCrtC (solid line) compared with the blank reaction (dotted line) and the pET15b-plasmid (dashed line). Crude extracts were assayed with 20 μM geranylgeraniol in 50 mM Na2HPO4 sodium phosphate (pH 8.0) at 30°C and 800 rpm. Peak 1, reaction product; Peak 2, geranylgeraniol.

In order to investigate how the newly identified key residues could be involved in the catalytic hydration reaction, the modeled structure of CrtC (Figure 3.4) was re-analyzed. We have assumed earlier that regions IV and V might contain potentially important residues, and indeed, it appears that our assumption is correct with regard to region IV. The identified key residues H239, W241, Y266 and D268 in RgCrtC and the corresponding residues in TrCrtC are all in close distance to each other (Figure 3.5). These four residues, which are conserved throughout the CrtC family, are also found in the active site of squalene-hopene cyclase (SHC) [14]. SHC catalyzes the cyclization reaction of squalene to hopene as a major product (Figure 3.10). Hopanol is also formed to a minor extent. The proposed mechanism for cyclases is proton-triggerd polycyclization, whereby the intermediate carbocation is stabilized by aromatic amino acids.

77 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

Table 3.4 Relative amount of the reaction product obtained from the substrate geranylgeraniol. Reactions were performed with Escherichia coli crude extracts expressing R. gelatinosus and T. roseopersicina CrtC wildtype and mutants, respectively. The areas under the substrate (1) and the product (2) peaks (Figure 3.9) were set as 100% and were used as the measure of the relative activity. N.d. not detected.

Reaction product (%) Sample after 1 day after 2 days RgCrtC wt 0.37 1.00

RgCrtC trunc 0.47 0.82

RgCrtC H239A <0.05 <0.05 RgCrtC W241A <0.05 <0.05 RgCrtC H264A 0.08 0.11 RgCrtC Y266A <0.05 <0.05 RgCrtC D268A <0.05 <0.05

TrCrtC wt 0.06 0.09

TrCrtC trunc 0.14 0.44 TrCrtC S58V 0.14 0.29 TrCrtC S58Q 0.15 0.21 TrCrtC H237A <0.05 <0.05 TrCrtC W239A <0.05 <0.05

TrCrtC H262A 0.16 0.25

TrCrtC D266A <0.05 <0.05

Next to the stabilization role of the aromatic amino acids, they also create hydrophobic environment in order to prevent quenching of the cation by water. The cyclization cascade is terminated by a well-positioned enzymatic base. The formation of the side alcohol product suggests significant water accessibility at the termination region of the active site.

78 3.3 Results and discussion

Figure 3.10 Enzyme catalyzed cyclisation of squalene to hopene and hopanol.

The acidic residue aspartate (D376), which is located in the center of the active site in SHC, is the likely general acid responsible for protonating the C3 atom of the squalene substrate [14]. The acidity of D376 is enhanced by a connection to the side chain of Y495 through a water molecule. Because of the similarity of the initial protonation reactions of squalene and lycopene, we assume that the residues involved in catalysis will be alike in SHC and CrtC. This hypothesis is in agreement with our results obtained by mutagenesis study as well as the structure-based analysis. Therefore, we propose the following mechanism for CrtC. D268 is the catalytic acid that initiates the hydration of lycopene (Figure 3.11).

Figure 3.11 Proposed mechanism for the initial protonation during lycopene hydration.

79 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina

Upon diffusion of lycopene into the active site, it is required that the C2 atom of the substrate is positioned near the proton of D268 that putatively will be added to the substrate. In order to enhance the acidity of the catalytic D268 for olefin protonation, the amino acid is directly bonded to H239 and to Y266 through an ordered water molecule, similar to what has been proposed for SHC [14]. Mutation of one of these three amino acids leads to inactivation of enzymatic activity, thereby supporting our hypothesis that they are all directly involved in the hydration reaction of lycopene. In contrast to SHC, where premature quenching of the cationic intermediate by water or nucleophiles is prevented by well positioned aromatic amino acids, a water molecule is added to lycopene to yield the desired hydroxylated lycopene derivative. This suggests that the active site of CrtC has more water molecules present, so that the interaction between the substrate and solvent water molecules is more significant. The aromatic amino acid Trp266 may be involved in the stabilization of the intermediate carbocation. Interestingly, according to the model, the residues H264 (Rg) and H262 (Tr), respectively, which seem not to have any functional property, are located away (region V) from the potential active site residues. This observation further supports our hypothesis that the region IV is the active site of CrtC. On the other hand, the three conserved hydrophobic residues proline, phenylalanine and tyrosine in region VI might play the role of attracting the hydrophobic substrate and placing it in the right position. The mainly hydrophobic amino acids in region II, which are located close to potential active side residues in region IV, might play a role in stabilization of the substrate during catalytic activity.

3.4 Conclusion

The main purpose of this study was to investigate whether it is possible to get more understanding of the hydration mechanism of carotenoid 1,2-hydratases. The used approach was modeling of the 3D structure with the closest homologous protein with known 3D structure, and subsequently the generation of point mutants of potentially important amino acid residues. Overall results indicate that the 3D structure consist of a beta-barrel, which closes with itself to form a central cavity. The substrate binding site, which consists mainly of hydrophobic amino acid residues, is located at the top of the cavity, while at the bottom inside the cavity potentially catalytic residues H, D, Y and W

80 3.5 Acknowledgements

are located. The absence of activity upon individual substitution of these residues by an alanine supports their roles in the initial protonation. Although, the model with only 17% sequence identity to the template is not very reliable, it fits to the data obtained on the activities of the mutants, reassuring that the model is probably correct. From our findings it becomes clear that the complete structure of the enzymes, through crystallization studies, will be pivotal to further unravel the mechanism for this intriguing enzyme. Nevertheless, the results of this study shed for the first time light on structure- activity relationships and opens the field for the engineering of carotenoid 1,2-hydratase to generate industrially relevant mutants.

3.5 Acknowledgements

We thank Prof. Dr. Jaap Jongejan for advice on the potentially important residues in carotenoid 1,2-hydratases and Jan van Leeuwen for making the CrtC model.

81 Structural Characterization of the Carotenoid 1,2-Hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina 3.6 References

[1] G. A. Armstrong and J. E. Hearst, "Carotenoids .2. Genetics and molecular biology of carotenoid pigment biosynthesis," Faseb Journal, vol. 10, pp. 228-237, Feb 1996. [2] S. L. Jensen, et al., "Biosynthesis of carotenoids in purple bacteria : A re-evaluation based on considerations of chemical structure," Nature, vol. 192, pp. 1168-1172, 1961. [3] C. I. Cazzonelli, "Carotenoids in nature: insights from plants and beyond," Functional Plant Biology, vol. 38, pp. 833-847, 2011. [4] G. A. Armstrong, "Genetics of eubacterial carotenoid biosynthesis: A colorful tale," Annual Review of Microbiology, vol. 51, pp. 629-659, 1997. [5] D. Umeno, et al., "Diversifying carotenoid biosynthetic pathways by directed evolution," Microbiology and Molecular Biology Reviews, vol. 69, pp. 51-78, Mar 2005. [6] Z. T. Sun, et al., "A novel carotenoid 1,2-hydratase (CruF) from two species of the non- photosynthetic bacterium Deinococcus," Microbiology-Sgm, vol. 155, pp. 2775-2783, Aug 2009. [7] A. Hiseni, et al., "Biochemical characterization of the carotenoid 1,2-hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina," Applied Microbiology and Biotechnology, vol. 91, pp. 1029-1036, 2011. [8] R. Kourist, et al., "Enzymatic synthesis of optically active tertiary alcohols: Expanding the biocatalysis toolbox," Chembiochem, vol. 9, pp. 491-498, Mar 3 2008. [9] J. F. Jin and U. Hanefeld, "The selective addition of water to C=C bonds; enzymes are the best chemists," Chemical Communications, vol. 47, pp. 2502-2510, 2011. [10] C. M. Evans and A. J. Kirby, "A model for olefin hydration: Intramolecular nucleophilic addition of phenolate oxygen to the unactivated double bond," Journal of the Chemical Society, Perkin Transactions 2, pp. 1259-1267, 1984. [11] H. J. Chiu, et al., "Structure of the first representative of Pfam family PF09410 (DUF2006) reveals a structural signature of the calycin superfamily that suggests a role in lipid metabolism," Acta Crystallographica Section F: Structural Biology and Crystallization Communications, vol. 66, pp. 1153-1159, 2010. [12] N. J. Patel, et al., "Use of deuterium labelling from deuterium oxide to demonstrate carotenoid transformations in photosynthetic bacteria," BBA - General Subjects, vol. 760, pp. 92-96, 1983. [13] A. A. Yeliseev and S. Kaplan, "Anaerobic carotenoid biosynthesis in Rhodobacter sphaeroides

2.4.1: H2O is a source of oxygen for the 1-methoxy group of spheroidene but not for the 2-oxo group of spheroidenone," Febs Letters, vol. 403, pp. 10-14, 1997. [14] K. U. Wendt, et al., "Enzyme Mechanisms for Polycyclic Triterpene Formation," Angewandte Chemie International Edition, vol. 39, pp. 2812-2833, 2000. [15] S. C. Hammer, et al., "Squalene hopene cyclases: highly promiscuous and evolvable catalysts for stereoselective CC and CX bond formation," Current Opinion in Chemical Biology, vol. 17, pp. 293- 300, 2013.

82 3.6 References

[16] S. F. Altschul, et al., "Basic Local Alignment Search Tool," Journal of Molecular Biology, vol. 215, pp. 403-410, Oct 5 1990. [17] A. Marchler-Bauer, et al., "CDD: A Conserved Domain Database for the functional annotation of proteins," Nucleic Acids Research, vol. 39, pp. D225-D229, 2011. [18] R. D. Finn, et al., "The Pfam protein families database," Nucleic Acids Research, vol. 38, pp. D211- D222, January 1, 2010 2010. [19] A. Dereeper, et al., "Phylogeny.fr: robust phylogenetic analysis for the non-specialist," Nucleic Acids Research, vol. 36, pp. W465-469, 2008. [20] A. Dereeper, et al., "BLAST-EXPLORER helps you building datasets for phylogenetic analysis," BMC Evolutionary Biology, vol. 10, 2010. [21] A. A. Porollo, et al., "POLYVIEW: A flexible visualization tool for structural and functional annotations of proteins," Bioinformatics, vol. 20, pp. 2460-2462, 2004. [22] K. Arnold, et al., "The SWISS-MODEL workspace: A web-based environment for protein structure homology modelling," Bioinformatics, vol. 22, pp. 195-201, 2006. [23] L. A. Waidner and D. L. Kirchman, "Aerobic anoxygenic photosynthesis genes and operons in uncultured bacteria in the Delaware River," Environmental Microbiology, vol. 7, pp. 1896-1908, Dec 2005. [24] E. Giraud, et al., "Two distinct crt gene clusters for two different functional classes of carotenoid in Bradyrhizobium," Journal of Biological Chemistry, vol. 279, pp. 15076-15083, Apr 9 2004. [25] M. Kammann, et al., "Rapid insertional mutagenesis of DNA by polymerase chain reaction (PCR)," Nucleic Acids Research, vol. 17, p. 5404, 1989. [26] G. Sarkar and S. S. Sommer, "The 'megaprimer' method of site-directed mutagenesis," BioTechniques, vol. 8, pp. 404-407, 1990. [27] O. Landt, et al., "A general method for rapid site-directed mutagenesis using the polymerase chain reaction," Gene, vol. 96, pp. 125-128, 1990. [28] N. Igarashi, et al., "Horizontal transfer of the photosynthesis gene cluster and operon rearrangement in purple bacteria," Journal of Molecular Evolution, vol. 52, pp. 333-341, Apr 2001. [29] M. J. Zvelebil, et al., "Prediction of protein secondary structure and active sites using the alignment of homologous sequences," Journal of Molecular Biology, vol. 195, pp. 957-961, 1987. [30] D. R. Flower, "The lipocalin protein family: Structure and function," Biochemical Journal, vol. 318, pp. 1-14, 1996. [31] J. M. LaLonde, et al., "The up-and-down beta-barrel proteins," The FASEB Journal, vol. 8, pp. 1240- 7, December 1, 1994 1994. [32] G. J. Bartlett, et al., "Analysis of catalytic residues in enzyme active sites," Journal of Molecular Biology, vol. 324, pp. 105-121, 2002.

83

Chapter 4 4 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection

Aida Hiseni, Rosario Medici, Isabel W.C.E. Arends, Linda G. Otten

Biotechnol. J. Feb 2014; DOI: 10.1002/biot.201300412

Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection Abstract

A novel high-throughput screening assay for the detection of alcohols is developed by using oleate hydratase (OHase) from Elizabethkingia meningoseptica as the model enzyme. It allows for screening of mutant libraries generated by directed evolution techniques or other mutagenesis methods. The assay is based on the reaction between the alcohol and nitrous acid to form the corresponding alkyl nitrite and is valid for a broad range of alcohols, differing in size and solubility. Cyclic as well as acyclic unsaturated alkenes (substrates) and the corresponding alcohols (products) were tested and they showed sufficient discrimination for analysis. Lower detection limits were 1.5 to 3 mM with excellent Z- factors ranging from 0.70 – 0.91. Precision, linearity and plate uniformity were estimated with pure substrates, mixtures and enzymatic reactions.

86 4.1 Introduction

4.1 Introduction

Oleate hydratase (OHase) (EC 4.2.1.53) belongs to the group of hydro-lyase enzymes (EC 4.2.1), which cofactor independently catalyze the reversible addition of a water molecule to a substrate possessing a carbon-carbon double bond [1]. The 73 kDa OHase from Elizabethkingia meningoseptica has been characterized with respect to its biochemical properties [2] and has recently been immobilized as CLEA (cross-linked enzyme aggregate) [3]. The enzyme shows high amino acid sequence similarities with members of the Streptococcal 67 kDa myosin-cross-reactive antigen like family [4, 5]. A few group members of this family were shown to also exhibit fatty acid hydratase activity [5-7]. Knowledge of the hydratase could be of great importance for the industry, as the reaction product 10-hydroxystearic acid (10-HSA) is a high-added-value compound and can be used for the production of a large number of industrial products including resins, waxes, nylons, plastics, cosmetics and coatings [3]. Compared to the traditional acid-catalyzed water addition, the enzymatic reaction proceeds under very mild conditions and is stereo- and regioselective. Recently, attempts have been made to increase microbial 10-HSA production by using recombinantly expressed oleate hydratase from Stenotrophomonas maltophilia [6, 8]. In industrial biocatalytic processes, hydro-lyases are underrepresented and only a few group members are amenable to be used for industrial scale reactions, including nitrile hydratase for the production of acrylamide [9] and fumarase to produce malate [10] . This is mainly due to the low stability and/or catalytic activity of these enzymes. However, utilization of this class of enzymes as biocatalysts needs intensive study and optimization of enzyme properties, such as stability, specific activity and selectivity, beforehand. A powerful tool to address these issues is directed evolution, which has been used in the past decade to improve biocatalysts [11-14]. An advantage of this technique is that no need for knowledge about the structure-function relationship is required. Moreover, a large number of mutants with potentially improved and/or novel properties can be produced in a short time. However, a crucial step in any directed evolution experiment is the development of a high-throughput screening (HTS) assay, which allows rapid screening of a large number of variants within a reasonable timeframe [15]. In general, the assay has to be sensitive, easy to perform, robust and has to have high throughput.

87 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection

So far, the quantification of enzymatic activity of OHase has usually been determined by either GC [2, 5-8] or HPLC [3, 16]. The methods are based on derivatization of fatty acids and are straightforward and accurate. Nevertheless, they are not suitable for high- throughput screening. Therefore, we have developed a spectrophotometric high-throughput screening method for the detection of alcohols. The assay is based on the simple reaction between the alcohol that is produced by the enzyme and nitrous acid to form the corresponding alkyl nitrite (Figure 4.1) [17].

Figure 4.1 Simplified scheme for the conversion of an alcohol to the corresponding ester of nitrous acid (box).

0,10 primary secondary 0,08 tertiary

0,06

0,04 Absorbance

0,02

0,00

330 340 350 360 370 380 390 400 410 420 430 440 Wavelength [nm]

Figure 4.2 Absorption spectra of 2-methylbutyl nitrite (primary), 3-methylbutan-2-yl nitrite (secondary) and tert-pentyl nitrite (tertiary) obtained from 2-methyl-1-butanol, 3-methyl-2-butanol and 2-methyl-2-butanol, respectively. 10 μl of the 150 mM stock in acetone was added to 90 μl of 20 mM Tris-HCl (pH 8.0) and the nitrosation reaction performed under standard conditions (as described in Materials and Methods, section “Assay conditions”).

88 4.2 Materials and methods

Next to enzyme activity detection, this assay provides also information about the position of the alcohol group in the molecule. For example, alkyl nitrites obtained from tertiary alcohols have a maximum at 400 nm, which is absent in those obtained from primary and secondary alcohols (Figure 4.2) [18]. With this valuable information the regioselectivity of an enzyme can be easily determined, especially of importance if the substrate contains several double bonds that could be converted by the same hydratase. The capability of the developed method for enabling an automated set up has been examined and characterized with OHase as the model enzyme. Next to precision of the system, the linearity and quality have also been addressed by using simulated and enzymatic reaction systems.

4.2 Materials and methods

4.2.1 Standard curves and Z-factor determination In order to validate the assay for linearity, standard curves and reaction simulation curves (substrate-product mixtures) were prepared using different alkenes and alcohols (Sigma- Aldrich) in a concentration range of 1 – 15 mM in triplicate. A desired amount of the 150 mM stock in acetone was manually added to the plates pre-filled with 20 mM Tris-HCl (pH 8.0). Subsequent steps were performed using the automated liquid handling system as described in “Assay conditions” section. In the case of fatty acids 12-hydroxystearic acid (12-HAS) and oleic acid (OA) (Sigma-Aldrich) a stock in DMSO was used and dispensed by Hamilton syringe for better accuracy. For the determination of the Z′-factor, a parameter that indicates the suitability of the assay to be used in high-throughput format, 15 mM alkene, alcohol and fatty acids, respectively, were used. Therefore, 10 μl of the 150 mM stock was added to plates pre-filled with 20 mM Tris-HCl (pH 8.0) (n = 32), and the nitrosation assay was performed using the standard conditions. The calculation of the Z′-factor was done using following equation: (3SD 3 SD ) Z1  alcohol alkene (Eq. 4.1) meanalcohol mean alkene To test the real case situation, E. coli TOP10 cells with pBAD-HISA-OH (pBAD/HisA vector containing ohyA gene) were used instead of the standards, whereby pBAD-HISA

89 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection

(pBAD/HisA empty vector) served as the negative control (n = 48). Following equation was used for the Z-factor calculation: (3SD 3 SD ) Z1 pBAD HISA OH pBAD  HISA (Eq. 4.2) meanpBAD HISA OH mean pBAD  HISA

4.2.2 Large scale production of 10-HSA As the OHase reaction product, 10-hydroxystearic acid (10-HSA), is commercially not available, we used 12-HSA instead, with the assumption that both have similar properties. In order to test this hypothesis, large scale production of 10-HSA was performed using E. coli TOP10 cells overexpressing OHase. The reaction mixture contained 1 ml of the cell- free extract (24 mg ml-1), 0.6 % (v/v) oleic acid in a total volume of 25 ml in 20 mM Tris- HCl (pH 8.0). After overnight incubation at 30°C and 200 rpm, a 500 μl sample was used to confirm the product by HPLC. First, the pH of the sample was adjusted with 50 μl of 3N HCl. Subsequently, 50 μl of saturated NaCl solution was added and fatty acids extracted with one volume of dichloromethane. The mixture was shaken for 5 min at 1400 rpm, centrifuged for 1 min at 13.200 rpm and 1 μl of the dichloromethane phase was transferred into a new tube. After drying with a SpeedVac Concentrator (Thermo Scientific), fatty acids were derivatized and analyzed as described earlier [3]. Once the reaction product was confirmed as 10-HSA, it was isolated as described previously [16] with small modification. Briefly, the rest of the mixture (24.5 ml) was filtered through Whatman filter-paper (Grade 1) and the filter with the residues dried overnight at 39°C. The dry solid was scraped off and dissolved in 15 ml of EtOH. Unsoluble material was filtered off and the filtrate dried using a SpeedVac Concentrator (Thermo). The isolated product was stored at 4°C until further use.

4.2.3 Growth conditions in 96-well deep well plates Well separated E. coli TOP10 colonies containing the plasmid pBAD-HISA-OH [3] or empty pBAD-HISA (control) were picked and transferred to individual wells in 96-well microtiter plates containing 150 μl LB medium with 100 μg ml-1 ampicillin and 0.2 % L- arabinose (w/v) followed by overnight incubation at 37°C and 150 rpm. Using a 96-pin colony replicator, the cells were transferred into 96-well deep well plates (2 ml, V-bottom, Greiner Bio-One), which were pre-filled with 1 ml of TB-medium and 100 μg ml-1 ampicillin. Prior to incubation for 24 h at 37°C and 150 rpm, the plates were covered with gas-permeable seals (BreathSeal, Greiner Bio-One) and lids (Lid with condensation ring,

90 4.2 Materials and methods

Greiner Bio-One). After harvesting by centrifugation for 35 min at 4000 rpm and 4°C (5810R, Eppendorf), the resulting cells were washed ones with 20 mM Tris-HCl (pH 8.0) and stored at -20°C overnight.

4.2.4 Liquid handling All liquid handling steps were performed using the JANUS® automated workstation (Perkin Elmer). The system is equipped with an 8-tip pipetting arm and an arm with a 96- channel dispense head.

4.2.5 Assay conditions Cell lysates were prepared by resuspending the cell pellets in 130 μl of 20 mM Tris-HCl (pH 8.0) containing 1 mg ml-1 lysozyme and 0.1 mg ml-1 DNAse and subsequent incubation at 37°C for 1 h. Cell debris and unlysed cells were removed by centrifugation for 35 min at 4000 rpm and 4°C. For enzymatic reactions, 95 μl of the cell-free extract was transferred into a 96-well deep well assay plate (1 ml, U-bottom, Greiner Bio-One) pre-filled with 5 μl of the substrate oleic acid (1 M stock in DMSO). The plates were covered and incubated overnight at 30°C and 300 rpm. For whole cell reactions, cell pellets were resuspended in 130 μl of 20 mM Tris-HCl (pH 8.0) and 95 μl of the cell suspension was directly transferred into the assay plate containing the substrate. The reaction mixtures were incubated overnight at 30°C and 300 rpm. Following steps were performed with the automated workstation (Figure 4.3).

Figure 4.3 Conceptualized 96-well high-throughput screening assay procedure using JANUS® automated workstation. Steps after the performance of the enzymatic reaction are shown. Refer to text for a stepwise description of the operation.

91 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection

First, 150 μl of octane was added to the assay plate (Figure 4.3, a), followed by 50 μl of

10% HCl (aq., v/v) (Figure 4.3, b) and 80 μl 10% NaNO2 (aq., w/v) (Figure 4.3, c) with the 96-channel dispense head. The plate was manually capped (CapMat, Greiner Bio-One), shaken vigorously for 3 min at 1500 rpm using a small shaker with microtiter attachment (IKA) and centrifuged for 3 min at 4000 rpm and RT. The plate was returned to the workstation and by using the 8-tip pipetting arm, 50 μl of the organic layer was transferred to a clean 96-well microtiter plate containing 50 μl of octane (Figure 4.3, d). The pipetting height was carefully chosen so that only organic layer was transferred. After a short agitation, absorption spectrum was taken between 330 nm and 440 nm using a spectrophotometer (Synergy 2, Bio Tek Instruments, Inc.). To calculate activity the absorbance of the highest peak was taken; this is approximately 356 nm, 372 nm and 382 nm for 1°, 2° and 3° alcohols respectively.

4.2.6 Preparation of ohyA mutant libraries For the generation of randomly mutated ohyA variants, epPCR (error-prone PCR) was performed across the entire coding sequence (1941 bp) using the Genemorph II random mutagenesis kit (Stratagene). Two libraries were constructed with low (0 – 4.5 mutations/kb) and high (9 – 16 mutations/kb) mutation frequencies. This could be achieved by adjusting the template DNA concentration. Amplification reactions were done under standard PCR conditions using plasmid pBAD-HISA-OH [3] as template and flanking primers pBAD_For (CTCTTCTCGCTAACCAAACC) and pBAD_Rev (GGCGTTTCACTTCGGCATGG). Prior to the cloning of the variant genes into an appropriate vector, the first PCR products were subjected to a second PCR reaction, where restriction sites for XhoI (forward) and HindIII (reverse) were introduced through a second set of primers (OH_F AATCTCGAGATGAACCCAATAACTTC; OH_R ATTAAGCTTTTATCCTCTTATTCCTTTTAC) (restriction sites are underlined) and the amount of the DNA was increased. Taq PCR master kit (Qiagen) was used following the manufacturer’s instruction. Using XhoI / HindIII restriction sites, the digested and purified fragments were ligated into the same sites of the pBAD-HISA vector and transformed into electrocompetent E. coli TOP10 cells. The insertion of the genes was verified by restriction analysis with XhoI / HindIII enzymes. From each library, a representative number of randomly selected clones was analyzed by sequencing (BaseClear).

92 4.2 Materials and methods

4.2.7 Expression of ohyA variants After the transformation of the recombinant plasmids into E. coli TOP10, the cells were spread on a 200 ml LB agar plate containing 100 μg ml-1 ampicillin and the plates were incubated at 37°C overnight. The volume of the cells used was adjusted so that between 1000 and 2000 clones would be present per plate. Individual colonies from mutant libraries were inoculated into individual wells of 96-well microtiter plate containing 150 μl LB medium supplemented with 100 μg ml-1 ampicillin with a VersArray Colony Picker (Biorad). By removing six of the colony picking needles, the empty wells could be manually inoculated with six wild type ohyA clones, which served as positive controls in each plate. The plates were covered with gaspermeable seals (BreathSeal, Greiner Bio- One) and incubated at 37°C overnight with shaking (150 rpm). Following this, the cells were transferred into 96-well deep well plates (2 ml, V-bottom, Greiner Bio-One) pre-filled with 1 ml of TB-medium and 100 μg ml-1 ampicillin using a colony copier. The rest of the cells was mixed with 40 μl of 60% glycerol (reference plate), sealed with seals (SilverSeal, Greiner Bio-One) and stored at -80°C. Prior to incubation for 24 h at 37°C and 150 rpm, the plates were covered with gaspermeable seals and lids (Lid with condensation ring, Greiner Bio-One). Next, protein expression was induced by adding 10 μl of a 20% L- arabinose stock using the JANUS® automated workstation (Perkin Elmer) and the plates incubated for another 24 h under the same conditions. After harvesting by centrifugation for 35 min at 4000 rpm and 4°C, the resulting cells were washed ones with 20 mM Tris- HCl (pH 8.0) and stored at -20°C overnight.

4.2.8 Library screening Cell pellets were resuspended in 130 μl of 20 mM Tris-HCl (pH 8.0). For enzymatic reactions, 95 μl of the cell suspension was transferred into a 96-well deep well assay plate (1 ml, U-bottom, Greiner Bio-One) pre-filled with 5 μl of the substrate oleic acid (1 M stock in DMSO). The plates were covered and the reaction mixtures incubated overnight at 30°C and 300 rpm. Following steps were performed with the automated workstation as described in the “Assay conditions” section

93 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection 4.3 Results and discussion

A high-throughput screening assay for hydro-lyases was developed. At first, a product with the alcohol group is obtained by the enzymatic reaction. Subsequently, the nitrosation reaction is performed and the resulting alkyl nitrite extracted using an immiscible organic solvent. Direct measurements of the UV spectra were used for product detection. After validation of the assay it was assessed using oleate hydratase from E. meningoseptica as the model enzyme.

4.3.1 Method performance and linearity with small substrates In order to demonstrate that the assay is applicable to molecules with different sizes, structures and substituents, and to test the precision and detection limits of the developed method, different concentrations of unsaturated cyclic and acyclic substrates and the corresponding alcohols were assessed under high-throughput assay conditions (100 μl). Nitrosation of alcohols is a standard procedure in laboratories for the synthesis of alkyl nitrites and some nitrosation reactions have also been adapted to the industrial scale [19]. However, our objective is the miniaturization of this reaction in order to be applicable to the 96-well plate format. To accommodate good mixing of the reaction mixture, 96-well deep well plates (1 ml) were used instead of 96-well microtiter plates (~0.3 ml). Standard curves ranging from 1 to 15 mM were constructed and the results (Figure 4.4) demonstrate that the discrimination between the reaction with the alkene (substrate) or the alcohol (product) is sufficient for analysis. However, a slight background reaction of the substrate was observed for the cyclic compounds. Lower limits of detection for each compound were 1.5 mM, except for compounds 2-methyl-1-butanol and 2-methyl-2- butanol, where the response below 3 mM was not linear or reliable with relative standard deviation (RSD) exceeding 20% (calculated based on three replicates). Thus, this assay is not applicable to reactions where low or new enzyme activities are to be discovered. On the other hand, the linearity of the standard curves from 3 to 15 mM indicates that the method can also be used for quantification. This is a remarkable result, since the assay has many steps and additional liquid-liquid extraction, which in general give high errors in 100 μl format. Moreover, these results demonstrate the high robustness of the assay despite the complex set-up.

94 4.3 Results and discussion

0,10 0,10 0,10 0,08 0,08 0,08 HO

0,06 0,06 HO 0,06

0,04 0,04 0,04 OH 0,02 0,02 0,02 0,00 0,00 0,00 Absorbance 356 nm 356 Absorbance 0 2 4 6 8 10121416 nm 382 Absorbance 0246810121416 nm 372 Absorbance 0246810121416 Conc. [mM] Conc. [mM] Conc. [mM]

0,10 0,10 0,10 HO 0,08 0,08 HO 0,08 OH 0,06 0,06 0,06 0,04 0,04 0,04 0,02 0,02 0,02 0,00 0,00 0,00 Absorbance 370 nm 370 Absorbance Absorbance 382 nm 382 Absorbance Absorbance 356 nm 356 Absorbance 0 2 4 6 8 10121416 0246810121416 0246810121416 Conc. [mM] Conc. [mM] Conc. [mM]

0,10 0,10 HO OH 0,08 0,08 0,06 0,06 0,04 0,04 0,02 0,02 0,00 0,00 Absorbance 372 nm 372 Absorbance 0246810121416 nm 360 Absorbance 0 2 4 6 8 10 12 14 16 Conc. [mM] Conc. [mM]

Figure 4.4 Analysis of small alkene/alcohol pairs using the proposed high throughput screening method. Nitrosation reactions were performed with different concentrations of the standard (stock in acetone) in 20 mM Tris-HCl (pH 8.0). Absorbance of the highest peak in the alcohol spectrum was plotted.

One important parameter in the described method is the choice of the organic solvent, which has to be screened depending on the expected reaction product. Several organic solvents were tested including octane, decane, dodecane, tetradecane and hexadecane. Octane showed the best results for compounds used in this study (data not shown).

4.3.2 Method performance for larger substrates and reaction simulation Besides small cyclic and acyclic unsaturated substrates, the larger substrate oleic acid was tested, differing significantly in size and solubility. As 10-HSA is not commercially available, 12-HSA was used in this study to identify and evaluate the OHase reaction product, which has the alcohol group at a different position (position 12 vs. 10). To determine if both products would react similarly in the proposed assay, a larger scale production of 10-HSA was performed with OHase. The product 10-HSA was isolated and confirmed with HPLC as > 85% pure. Nitrosation experiments were performed with 5 mM

95 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection

10- and 12-HSA and the results are depicted in Figure 4.5. They clearly show that 12-HSA can be used as a standard for 10-HSA, since the spectra are identical. Lower purity of 10- HSA explains the slightly lower values in the absorption spectra with the same concentration.

0,030 10HSA 12HSA 0,025 OA

0,020

0,015

Absorbance 0,010

0,005

0,000 330 340 350 360 370 380 390 400 410 420 430 440 Wavelength [nm]

Figure 4.5 Absorption spectra of 10-(nitrosooxy)octadecanoic acid obtained from 10-HSA, 12- (nitrosooxy)octadecanoic acid obtained from 12-HSA and control reaction with OA. 50 mM stock in DMSO was diluted 10 x with 20 mM Tris-HCl (pH 8.0) and the nitrosation reaction performed using standard conditions.

Nitrosation of substrate oleic acid and the (artificial) product 12-HSA were performed. As shown in Figure 4.6, the obtained standard curve illustrates that even with very unsoluble substrates / products this HTS method can be used. No background reaction has been observed from the substrate oleic acid. Thus, the assay can be applied for the enzymatic reaction with OHase. Next, enzymatic reaction progress was simulated by using different mixtures of the substrate oleic acid and the product 12-HSA. The performed experiment gives an indication of the approach applicability to real situation, where different ratios of substrate and product are present in the mixture at the same time. From the results presented in Figure 4.7, a clear overlap of the simulated and the standard curve was observed, showing that the assay is not negatively influenced by the presence of the substrate during the assay.

96 4.3 Results and discussion

0,08

0,06

0,04

Absorbance 372 nm Absorbance 0,02

0,00 02468101214 Concentration [mM]

Figure 4.6 Analysis of 12-HSA/OA pair using the proposed high throughput screening method. Nitrosation reactions were performed with different concentrations of the standard (stock in DMSO) in 20 mM Tris-HCl (pH 8.0).

0,08

0,06

0,04

Absorption 372 nm 372 Absorption 0,02

Reaction simulation 12-HSA 0,00 0 2 4 6 8 10 12 14 Concentration [mM]

Figure 4.7 Simulated reaction progress and standard curve of 12-(nitrosooxy)octadecanoic acid obtained from 12-HSA. Different ratios of OA and 12-HSA were prepared (total concentration 16 mM) to simulate enzymatic reaction. In comparison, the standard curve of nitrosation of 12-HSA only, was plotted (Figure 4.6).

97 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection 4.3.3 Precision and accuracy (Z-factor) To develop a HTS method that can be used for the identification of ‘hits’, robustness and reproducibility are of a big importance. Therefore, a simple parameter (Z-factor) was defined by Zhang et al. [20], which can be used to evaluate the quality of the HTS method and recently has been applied for the detection of decarboxylase activities [21]. As this factor is dimensionless, it can be used to compare different HTS methods. The factor takes into account standard deviations and mean values of sample and control measurements (Eq. 4.1). In this study, first the Z′-factor was determined for all tested compounds to evaluate the quality of the assay itself and a summary of the results is shown in Table 4.1.

Table 4.1 Z′-factor values of various alkene / alcohol pairs, indicating the screening assay quality. The Z′-

factor was calculated using the formula (Eq. 4.1): Z′-factor = 1 – 3 × (SDalcohol + SDalkene) / |meanalcohol –

meanalkene| Alkene / Alcohol Z′-factor 2-methyl-1-butene / 2-methyl-1-butanol 0.82 2-methyl-1-butene / 2-methyl-2-butanol 0.71 2-methyl-2-butene / 2-methyl-2-butanol 0.70 2-methyl-2-butene / 3-methyl-2-butanol 0.72 2-methyl-1-pentene / 2-methyl-1-pentanol 0.91 2-methyl-1-pentene / 2-methyl-2-pentanol 0.73 2-methyl-2-pentene / 2-methyl-2-pentanol 0.73 2-methyl-2-pentene / 2-methyl-3-pentanol 0.90 1-methylcyclohexene / 2-methylcyclohexanol 0.72 3-methylcyclohexene / 3-methylcyclohexanol 0.78 oleic acid / 12-hydroxystearic acid 0.75

The Z′-factors are all >0.5, while most of them are even >0.7. Z′-factor values ranging from 0.5 to 1 demonstrate a good quality of the assay and can be used for identification of ‘hits’ [20]. Thus, the obtained values indicate that the HTS assay is an ‘excellent assay’ for all alkene/alcohol pairs tested, which makes this assay applicable to distinguish all kind of hydroxylated compounds from their unsaturated counterparts in only one step. Another important parameter when developing a 96-well format assay is the reproducibility and spatial plate uniformity. The design of the plates and the process can introduce new problems, for example, temperature gradients and aeration differences, especially when plates are stacked during the incubation time. The results can reveal patterns of drifts or edge effects, which have to be diminished by design optimization [22]. For this reason, a

98 4.3 Results and discussion scatter plot was prepared by plotting the response of the standards OA and 12-HSA against well number (n = 32), either by column or by row. The results (Figure 4.8A, B) show that no significant drifts were observed for the fatty acid standards, neither when plotting by column nor when plotting by row. This indicates that the process design of the assay is appropriate for high throughput screening. From these results we also can conclude that the automated liquid handling workstation works with high accuracy.

(A) 0,08 12-HSA OA

0,06

0,04

0,02 Absorbance 372 nm Absorbance

0,00

0 1224364860728496 Well number, by row 0,08 12-HSA (B) OA

0,06

0,04

0,02 Absorbance 372 nm 372 Absorbance

0,00

0 8 16 24 32 40 48 56 64 72 80 88 96 Well number, cy column

Figure 4.8 Scatter plot to assess the spatial uniformity. The response of nitrosated 15 mM 12-HAS and 15 mM OA is plotted against well number, ordered by row (A) and by column (B). The nitrosation reaction was performed under standard conditions.

99 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection 4.3.4 Optimization of protein expression conditions In order to test the applicability of the assay with real samples, E. coli TOP10 cell-extracts and whole cells with and without overexpressed OHase were used. This is especially important, as cell-extracts or whole cells naturally contain a lot of alcohols, which can serve as substrate for the nitrosation reaction and may result in increased background signal. In an effort to reduce the background signal, initial optimization of the expression was performed by using TB-medium (Terrific Broth) instead of the usual LB-medium for the enzyme expression. As expected, 2-fold higher protein concentrations were obtained with TB-medium compared to the LB-medium (data not shown). Furthermore, reaction with whole cells showed higher response to that obtained from cell-free extracts under the same reaction conditions and was chosen for further experiments to simplify the HTS process. 10-HSA production was performed in 96-well plates using E. coli TOP10 cells carrying either empty an expression vector (negative control) or a vector with ohyA gene (positive control). Under the chosen assay conditions (50 mM oleic acid, 30°C, overnight incubation) around 30% of the oleic acid was converted to 10-HSA (data not shown). Only partial conversion of the substrate is required in order to be able to detect potentially improved activity as long as the amount produced is above the detection limit. The nitrosation of the reaction products was conducted using the automated workstation. In this experiment, not only the accuracy of the workstation was tested, but also the spatial uniformity of the cell growth, substrate addition and temperature influence. The obtained results (data not shown) revealed substantial deviations in the response. In order to identify factors, which are affecting these results, further experiments were performed. At first, we have looked into the growth behavior of bacteria in 96-well deep well plates as it can differ from that in standard laboratory containers like Erlenmeyer flasks. The cell densities showed significant deviations during the first 10 hours of the growth (Figure 4.9). Usually, the protein expression is induced in the exponential phase of the growth, when cell densities reach a value of ~0.4 (OD600). However, with the miniaturization of cell cultures like in the high- throughput screening set-up, this would decrease the efficiency because of the added step in the procedure. Hence, the inducer agent is added immediately at the beginning of the cell growth. Since plates are inoculated with a colony replicator, cell densities are relatively different at that point. Consequently, this results in different enzyme quantities in each well. In order to have approximately the same amount of the enzyme in each well, the protocol

100 4.3 Results and discussion was adapted and the protein expression was induced after 25 h growth, where the standard deviation showed an acceptable value of 6.3% (Figure 4.9).

0,8 Cell density E. coli TOP10 0,7

0,6

0,5

600 0,4 OD 0,3

Time (h) Mean SD SD (%) 0,2 4 0,01 0,003 29,5 8 0,2 0,026 13,4

0,1 10 0,42 0,062 14,6 25 0,67 0,042 6,3 0,0 0 5 10 15 20 25 30 Time [h]

Figure 4.9 Growth curve of E. coli TOP10 in 96-well deep well plates. Mean values and standard deviations of the cell densities from 96 E. coli cultures are shown, which were grown at 150 rpm and 37°C over a period of 25 h.

The enzymatic reaction was repeated with cells grown under optimized conditions. Despite inducing OHase production later, a high standard deviation was obtained in the response. In an attempt to decrease the error in the OHase response, the protein induction time was varied between 16 and 25 h. As shown in Table 4.2, the Z-factor increased significantly from -46.50 to -0.77 by increasing the induction time.

Table 4.2 Z-factor values of E. coli cultures containing pBAD-HISA or pBAD-HISA-OH plasmid, indicating the total screening assay quality. The Z-factor was calculated using the formula (Eq. 4.2): Z-factor = 1 – 3 ×

(SDpBAD-HISA-OH + SDpBAD-HISA) / |meanpBAD-HISA-OH – meanpBAD-HISA|. Experiment Induction time (h) Z-factor 1 16 -46.50 2 21 -1.0 3 25 -0.77

101 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection

In addition, responses obtained were assessed for patterns of drift or edge effects, by plotting against well number and ordered either by row first, then by column (Figure 4.10A), or by column first, then by row (Figure 4.10B).

E. coli pBAD-HISA-OH (A) 0,10 E. coli pBAD-HISA 0,09 0,08 0,07 0,06 0,05 0,04 0,03

Absorbance 370 nm 0,02 0,01 0,00 -0,01 0 8 16 24 32 40 48 56 64 72 80 88 96 Well number, by column E. coli pBAD-HISA-OH 0,10 E. coli pBAD-HISA (B) 0,09 0,08 0,07 0,06 0,05 0,04 0,03

Absorbance 370 nm 0,02 0,01 0,00 -0,01 0 1224364860728496

Well number, by row

Figure 4.10 Scatter plot to assess the spatial uniformity. The response of nitrosated E. coli pBAD-HISA-OH and E. coli pBAD-HISA (n=48) reaction products is plotted against well number, ordered by row (A) and by column (B). The enzymatic and nitrosation reactions were performed under standard conditions.

102 4.3 Results and discussion

There are no drift patterns or edge affects visible in these figures, indicating that bacterial growth is not affected by the place it is situated. It is, however, obvious from the plots that the negative and positive points overlap, which will result in more false positives or missing of hits, depending on the threshold set. Although the deviation of the negative controls is larger than in the assay alone (32% vs 26%) it is still reasonable, considering the extra steps in this process. The RSD of 26% for the positive controls indicates a substantial variability in the response, since the 12-HSA variation itself was only 3.3%. This is probably due to variability in cell growth and protein production, despite the improvements already implemented. Another point to investigate is the octane extraction step. Compared to the used oleic acid and 12-HSA standards, whole cells consists of a complex mixture of different compounds, which may influence the extraction and introduce deviations to the assay. In order to be applicable for the detection of potential ‘hits’ from mutant library, the cell growth and enzyme production as well as the extraction step need further investigations. Adding extra steps will take more time and effort, but hopefully result in a good assay. Improvements should be at the beginning of the process, for instance, inoculation from pre-grown plates by the pipetting robot instead of the colony replicator tool, or afterwards by improving the extraction step. Also changing host or expression vector could improve enzyme variability. As changing the induction time already improved the Z-factor 60-fold, it should be possible to reduce the standard deviation of the OHase signal in the plate by another factor of 10, leading to a Z-factor >0.5, which would be an ideal assay. From the results described so far, it is clear that further improvements are needed in order to implement the new developed assay in the directed/random evolution studies. However, for the time being we applied the new high-throughput screening assay to identify an OHase variant with improved activity towards oleic acid by using a threshold value of 2 standard deviations of the wildtype OHase activity in each plate. The generation of the mutants included random mutagenesis of the ohyA, subsequent cloning in pBAD-HISA vector and expression in E.coli TOP10 cells. Two separate mutant libraries were prepared with either low- or high mutation frequency. The analysis of 55 randomly picked clones revealed an average mutation frequency of 5 mutations/kb, which is close to the expected range of 0 – 4.5 mutations/kb (data not shown). The average mutation frequency on amino acid sequence level was 3 mutations/gene. Approximately 1500 transformants were obtained in each library. In first instance, the low mutation frequency library was tested and the absorbances at 372 nm analyzed. An example

103 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection of the screening results is shown in Figure 4.11. The wild type OHase (Figure 4.11, black bars) was included in each plate to account for the variation of the activities.

0,10 0,08 0,06 0,04 0,02 Absorbance 372 nm 0,00 F1 F2 F3 F4 F5 F6 F7 F8 F9 E1 E2 E3 E4 E5 E6 E7 E8 E9 B1 B2 B3 B4 B5 B6 B7 B8 B9 C1 C2 C3 C4 C5 C6 C7 C8 C9 A1 A2 A3 A4 A5 A6 A7 A8 A9 D1 D2 D3 D4 D5 D6 D7 D8 D9 G1 G2 G3 G4 G5 G6 G7 G8 G9 H1 H2 H3 H4 H5 H7 H8 H9 F10 F11 F12 E10 E11 E12 B10 B11 B12 C10 C11 C12 A10 A11 A12 D10 D11 D12 G10 G11 G12 H10 H11

0,10 G12 0,08 wildtype average 0,06 0,04 Absorbance 0,02 0,00 340 360 380 400 420 440 Wavelength

Figure 4.11 Evaluation scheme of a screening process. To account for the variation of the activities of the individual variants (gray bars), a threshold (dashed line) was defined, above which a variant was declared as positive. The threshold was calculated for each plate using the data of the control (wildtype) activities (black bars). The absorbance spectra of a positive mutant and a wildtype are displayed.

13 mutants exhibiting higher activity than the wild type were identified by means of the colorimetric high-throughput screening method (Table 4.3). Due to the restricted time, further analysis of the obtained mutants could not be executed. However, in order to confirm the screening result, the obtained clones need to be re- cultivated and re-tested for activity. Once the results are re-produced and confirmed, more detailed analysis of the mutations in these particular mutants would be of high interest, as they could give an indication of directions for further improvement of the activity.

104 4.3 Results and discussion

Table 4.3 Screening analysis of 1500 epPCR OHase variants. The normalized values at λmax = 372nm of the OHase variants were compared to the values obtained for the wildtype OHase in the same plate. All showed mutants were declared as positive using the corresponding threshold values.

Absorbance 372 nm 96-well plate Threshold Wildtype number (wt average + 2SD) Mut1 Mut2 Mut3 Mut4 (average)

01 0.051 0.033 0.054 0.064 0.052

02 0.061 0.045 0.072 0.069 0.069 0.063

07 0.082 0.062 0.090

09 0.045 0.023 0.048 0.050

15 0.089 0.050 0.092 0.096 0.100

In conclusion, a robust, high quality high-throughput screening method has been developed for the detection of alcohols. In general, this assay can be used with any hydro-lyase member, whose product can undergo a reaction with a nitrosating agent to form alkyl nitrites. The assay is applicable to a broad range of compounds varying in size and solubility, with good to excellent Z′-factors. Future studies will be directed at optimization of the assay procedure for improvement of the plate uniformity of the enzyme concentration and of the octane extraction.

105 Oleate hydratase as model enzyme to design and evaluate high-throughput screening assay for alcohol detection 4.4 References

[1] J. F. Jin and U. Hanefeld, "The selective addition of water to C=C bonds; enzymes are the best chemists," Chemical Communications, vol. 47, pp. 2502-2510, 2011. [2] L. E. Bevers, et al., "Oleate hydratase catalyzes the hydration of a nonactivated carbon-carbon bond," Journal of Bacteriology, vol. 191, pp. 5010-5012, Aug 2009. [3] A. Hiseni, et al., "Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA)," 2012. [4] E. Rosberg-Cody, et al., "Myosin-cross-reactive antigen (MCRA) protein from Bifidobacterium breve is a FAD-dependent fatty acid hydratase which has a function in stress protection," Bmc Biochemistry, vol. 12, Feb 2011. [5] A. Volkov, et al., "Myosin cross-reactive antigen of Streptococcus pyogenes M49 encodes a fatty acid double bond hydratase that plays a role in oleic acid detoxification and bacterial virulence," Journal of Biological Chemistry, vol. 285, pp. 10353-10361, 2010. [6] Y.-C. Joo, et al., "Production of 10-hydroxystearic acid from oleic acid by whole cells of recombinant Escherichia coli containing oleate hydratase from Stenotrophomonas maltophilia," Journal of Biotechnology. [7] Y. C. Joo, et al., "Biochemical characterization and FAD-binding analysis of oleate hydratase from Macrococcus caseolyticus," Biochimie. [8] E. Y. Jeon, et al., "Bioprocess engineering to produce 10-hydroxystearic acid from oleic acid by recombinant Escherichia coli expressing the oleate hydratase gene of Stenotrophomonas maltophilia," Process Biochemistry, vol. 47, pp. 941-947, 2012. [9] S. van Pelt, et al., "Nitrile hydratase CLEAs: The immobilization and stabilization of an industrially important enzyme," Green Chemistry, vol. 10, pp. 395-400, 2008. [10] A. S. Bommarius and B. R. Riebel, "Introduction to Biocatalysis," in Biocatalysis, ed: Wiley-VCH Verlag GmbH & Co. KGaA, 2005, pp. 1-18. [11] F. H. Arnold, "Design by directed evolution," Accounts of Chemical Research, vol. 31, pp. 125-131, Mar 1998. [12] U. T. Bornscheuer and M. Pohl, "Improved biocatalysts by directed evolution and rational protein design," Current Opinion in Chemical Biology, vol. 5, pp. 137-143, Apr 2001. [13] L. G. Otten and W. J. Quax, "Directed evolution: selecting today's biocatalysts," Biomolecular Engineering, vol. 22, pp. 1-9, Jun 2005. [14] L. G. Otten, et al., "Enzyme engineering for enantioselectivity: from trial-and-error to rational design?," Trends in Biotechnology, vol. 28, pp. 46-54, Jan 2009. [15] J.-L. Reymond, Enzyme assays: High-throughput screening, genetic selection and fingerprinting: Whiley-VCH, 2006. [16] J. A. Hudson, et al., "Conversion of oleic acid to 10-hydroxystearic acid by two species of ruminal bacteria," Applied Microbiology and Biotechnology, vol. 44, pp. 1-6, Dec 1995.

106 4.4 References

[17] D. L. H. Williams, "O-Nitrosation," in Nitrosation Reactions and the Chemistry of , ed Amsterdam: Elsevier Science, 2004, pp. 105-115. [18] I. A. Leenson, "Identification of primary, secondary, and tertiary alcohols - An experiment in spectrophotometry, organic chemistry, and analytical chemistry," Journal of Chemical Education, vol. 74, pp. 424-425, Apr 1997. [19] W. Lyn, "Introduction," in Nitrosation Reactions and the Chemistry of Nitric Oxide, ed Amsterdam: Elsevier Science, 2004, pp. xi-xii. [20] J. H. Zhang, et al., "A simple statistical parameter for use in evaluation and validation of high throughput screening assays," Journal of Biomolecular Screening, vol. 4, pp. 67-73, Apr 1999. [21] R. Médici, et al., "A high-throughput screening assay for amino acid decarboxylase activity," Advanced Synthesis and Catalysis, vol. 353, pp. 2369-2376, 2011. [22] B. Eastwood, et al. (2009). Assay Guidance Manual, Version 6 [from internet]. Available: http://assay.nih.gov/assay/index.php/Table_og_Contents

107

Chapter 5

5 Preparation and properties of immobilized oleate hydratase as a cross- linked enzyme aggregate (CLEA)

Aida Hiseni, Maria del Rosario Franco Berriel, Isabel W.C.E. Arends and

Linda G. Otten

Manuscript in preparation

Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA) Abstract

The immobilization of oleate hydratase (OHase) from Elizabethkingia meningoseptica as cross-linked enzyme aggregates (CLEA) is described. CLEA’s were prepared by precipitation of OHase in cell-free extracts and subsequent cross-linking with glutaraldehyde. The effects of different precipitating agents and different concentrations of the cross-linking agent glutaraldehyde were investigated. In an optimized procedure ninety percent ammonium sulfate saturation and 0.3wt% glutaraldehyde were used. Activity recovery of 26% was achieved after 21 h cross-linking at 4°C. OHase CLEA’s had increased activity over a range of different temperatures compared to that of the OHase both in cell-free extracts as well as in whole cells.

110 5.1 Introduction

5.1 Introduction

The possibility of using renewable, plant-based resources for the production of fine chemicals has generated wide interest in the field of industrial biotechnology. Reasons such as fluctuating supply and price of the finite resource petroleum, and considerable environmental issues have led to the development of new processes in the industry, where petroleum-based products are being replaced by products derived from alternative and sustainable sources [1-4]. One of the cheapest and most abundant biological raw materials is vegetable oil [5]. Hydroxy fatty acids (HFA) have specific physical and chemical properties i.e. high viscosity and reactivity, which make them suitable for the production of a number of products, including resins, nylons, plastics, waxes, cosmetics and coatings [6]. They can be obtained by chemical modification of unsaturated fatty acids using strong acids such as sulfuric acid, followed by subsequent hydrolysis [6]. However, the resulting mixture of several HFA’s requires costly downstream processing. Furthermore, regio- and enantioselectivity is difficult to achieve. Therefore, the use of isolated enzymes and/or microbial systems will offer significant advantages: Both the problem of selectivity as well as the requirement of strong acids can be overcome. The enzymatic hydration of oleic acid (OA) into 10-hydroxystearic acid (10-HSA) was first described in a Pseudomonas strain [7]. Since then reports followed for a series of different bacterial and eukaryotic microorganisms, such as Sphingobacterium thalpophilum [8] Corynebacterium [9], Saccharomyces cerevisiae [6] and Stenotrophomonas nitritireducens [10] with space-time yields ranging from 0.001 to 16 g l-1 h-1 for 10-HSA (Supplementary table 5.1). Although over the years much research has been devoted to the optimization of the fermentation conditions so that high productivities of the enantiomerically pure 10- HSA can be obtained, rather little attention has been paid to the enzyme responsible for this hydration reaction. Only recently, Bevers et al. [11] were able to recombinantly express and characterize oleate hydratase (OHase) from Elizabethkingia meningoseptica (formerly known as Pseudomonas sp. 3266), the same strain that Davis et al. [12] described 43 years ago. This hydratase represents a new type of hydro-lyase, which is able to hydrate an isolated carbon-carbon double bond (Figure 5.1) and is a possible biocatalyst for the production of several alcohols and alkenes [13, 14]. The growing interest in this type of hydro-lyases has been shown by many recent studies that have focused on finding and

111 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA)

characterizing oleate hydratases from other microorganisms for the production of 10-HSA [15-18]. O

OH oleic acid

OHase H O 2 O

OH OH 10-hydroxystearic acid

Figure 5.1 Reaction catalyzed by oleate hydratase (OHase): conversion of oleic acid into 10-hydroxystearic acid.

The use of isolated enzymes in biocatalytic transformations allows for higher product concentrations, less side reactions and simplified down-stream processing compared to processes catalyzed by whole cells. Furthermore, diffusion limitations do not occur. On the other hand, isolated enzymes may show high sensitivity to industrial conditions, which often involve organic solvents, extreme pH’s or elevated temperatures. One strategy to improve operational performances of enzymes in industrial processes is immobilization as a means of stabilization. Furthermore, immobilization may improve other enzyme properties, including selectivity or specificity or reduce enzyme inhibition by, for instance, substrate or product [19]. A considerable number of immobilization techniques, such as binding to a carrier, encapsulation in a polymeric matrix or cross-linking of enzyme aggregates, are known to date [20-22]. However, there is no universal protocol. For every enzyme the best immobilization method needs to be investigated. Hanefeld et al. [23] and Garcia-Galan et al. [24] have reviewed enzyme immobilization and they highlighted some important parameters that have to be taken into account when choosing a suitable immobilization technique. To our knowledge, the immobilization of oleate hydratase has not been reported to date. As a first and straightforward methodology to immobilize OHase cross-linked enzyme aggregates (CLEA) were prepared. This method offers the advantage that it does not lead to ‘dilution of activity’ by usage of an carrier, has lower production costs through exclusion of an expensive carrier, and produces a catalyst with highly concentrated activity [25]. In this study, the results obtained for the immobilization of the overexpressed non-purified

112 5.2 Materials and Methods

OHase are described. For this purpose, recombinant OHase from E. coli cell-free extracts was aggregated and cross-linked using a bifunctional cross-linker glutaraldehyde. Biochemical and biophysical properties as well as the efficiency of the CLEA biocatalyst were investigated.

5.2 Materials and Methods

5.2.1 Bacterial strain, growth conditions and cell disruption E. coli TOP10 cells containing the plasmid pBAD-HISA-OH [11] were grown at 37°C in -1 TB medium with 100 μg ml ampicillin until an OD600 value of 0.6 – 0.8 was reached. Protein expression was induced with 0.2% arabinose (final concentration), followed by cultivation at 28°C overnight. Cells were harvested by centrifugation (10.000 rpm, 10 min, 4°C; Sorvall), washed once with 20 mM Tris-HCl pH 8.0 and lysed in the same buffer with a cell disruptor at the pressure of 1.5 kBar (Constant systems, IUL instruments). Cell-free extract (CFE) was separated from cell debris by centrifugation at 10.000 rpm for 20 min at 4°C and stored on ice until further use. For long term storage, aliquots of CFE were frozen in liquid nitrogen and stored at -80°C.

5.2.2 Precipitation procedure An amount of 0.1 ml of CFE (protein concentration: 4 mg ml-1) was added drop-wise to 0.9 ml precipitant (acetone, acetonitrile (ACN), ethanol, 2-propanol, 1,2-dimethoxyethane (DME) or saturated ammonium sulfate), at room temperature (RT) and 4°C, respectively. The resulting mixture was shaken at 400 rpm (Eppendorf Thermomixer) for 1 h, after which the precipitated protein was separated at 13.200 rpm for 20 min (Eppendorf centrifuge). Subsequently, the pellets were resuspended in 0.5 ml 20 mM Tris-HCl pH 8.0, and assayed for activity using oleic acid as substrate.

5.2.3 Cross-linking procedure After protein precipitation 12.5 – 200 μl of 25wt% glutaraldehyde was added drop-wise into the same tube and the mixture shaken at 400 rpm for 1 – 21 h. When the protein concentration was varied from 1 to 24 mg ml-1 0.3wt% of glutaraldehyde was used. The suspended CLEA’s were centrifuged (13.200 rpm, 30 min) and the supernatant was removed. In order to remove non-cross-linked protein and the remaining glutaraldehyde,

113 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA)

CLEA’s were washed three times with 0.5 ml 20 mM Tris-HCl pH 8.0 and stored in the same buffer on ice until further use. Washing supernatants were assayed for activity to determine enzyme leakage. For the scale-up procedure, the above described protocol was used, with 50-fold increased amount of OHase. Briefly, 5 ml of CFE (protein concentration: 4 mg ml-1) was added drop- wise to 45 ml saturated ammonium sulfate solution and the mixture shaken at 40 rpm (Incubator shaker, Innova 44) and 4°C for 1 h. Subsequently, 625 μl of 25wt% glutaraldehyde (endconc. 0.3wt%) was added and the mixture shaken further for 21 h. The CLEA’s were centrifuged for 30 min at 4000 rpm and the supernatant removed. After washing three times with 1 ml 20 mM Tris-HCl pH 8.0, the CLEA’s were stored in the same buffer on ice until further use.

5.2.4 Activity assay OHase activity was determined using oleic acid as substrate. Unless otherwise stated, a standard assay was performed with 0.5 ml final volume in 20 mM Tris-HCl pH 8.0, containing 6 mM oleic acid and 2 μl of cell-suspension (55 mg ml-1), 2 μl of CFE (24 mg ml-1) or a certain amount of CLEA, respectively. The mixtures were incubated at 30°C and 1400 rpm. After a desired time interval the reaction was stopped by the addition of 50 μl 3N hydrochloric acid (HCl), and substrate and product were extracted from the aqueous layer. Prior to the extraction stearic acid (dissolved in acetone) was added, which served as external standard, followed by addition of 50 μl saturated NaCl solution and the extraction with one volume of dichloromethane (DCM). The mixtures were shaken for 5 min at 1400 rpm, centrifuged for 1 min at 13.200 rpm and 50 μl of the DCM phase was transferred into a new tube. After drying with a SpeedVac Concentrator (Thermo), fatty acids were derivatized [26, 27]. To the dried extracted fatty acids 25 μl of 2-bromoacetophenone (10 mg ml-1 in acetone) and 25 μl of triethylamine (10 mg ml-1 in acetone) were added and the mixtures heated at 96°C for 15 min. 3.5 μl of acetic acid was added and the mixtures heated for a further 5 min. After evaporation to dryness the samples were reconstituted in 0.1 ml of ACN for HPLC analysis. Separation was performed with a 4.6 x 50 Merck Chromolith

SpeedROD RP-18e, using H2O-ACN mobile phase gradient (A: H2O with 0.1%, v/v trifluoroacetic acid; B: ACN). The gradient consisted of 50% B over 3 min, 50 – 80% B over 2 min and isocratic elution (80% B) over 7 min, at 1 ml min-1 and at a column temperature of 50°C. Derivatized substrate and product were detected at 242 nm (SPD20A, Shimadzu).

114 5.3 Results and discussion

For quantitative analysis a linear relationship was established for the peak area ratios of product versus external standard stearic acid.

5.2.5 Storage stability Storage stabilities of cell-suspension (55 mg ml-1), CFE (24 mg ml-1) and CLEA (0.2 mg ml-1) were tested by storing the enzyme in 20 mM Tris-HCl pH 8.0 at 4°C and RT (21°C), respectively, for several days. At various time points activities were determined using oleic acid as substrate under standard assay conditions. Stabilities were given as residual activities, calculated by taking the initial activity of the enzyme as 100%.

5.2.6 pH activity and temperature stability In order to investigate the pH effect on the enzyme activity, standard assay conditions were used in buffers with varying pH values (100 mM sodium acetate, pH 3.0 – 6.0; 100 mM potassium phosphate, pH 6.0 – 8.0; 50 mM Tris-HCl pH 8.0 – 9.0). Enzyme activity, determined in 20 mM Tris-HCl pH 8.0 under standard conditions was designated as 100%. Thermal stability was investigated by pre-incubating the enzyme at temperatures ranging from 20 to 50°C in the absence of substrate for 20 min, cooling the enzyme solution on ice, and then measuring the residual activity using the standard assay. Residual activities were calculated by taking the initial activity of the enzyme as 100%.

5.2.7 Biocatalyst recovery Biocatalyst operational stability was studied using standard assay containing 10 mg ml-1 of CLEA. After 1 h of reaction, product extraction and derivatization were performed as described in section “Activity assay”. CLEA was removed from the interphase after extraction, washed three times with 0.5 ml 20 mM Tris-HCl pH 8.0, and resuspended in fresh buffer to perform a new reaction.

5.3 Results and discussion

Recombinantly overexpressed OHase from E. meningoseptica was immobilized by self- aggregation into cross-linked enzyme aggregates (CLEA’s). The crude enzyme is first aggregated by a precipitating agent and subsequently covalently cross-linked using a bi- functional agent glutaraldehyde [20]. One advantage of this method is that this immobilization procedure is carrier-free. Protein sequence analysis of OHase revealed 50

115 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA)

lysine residues, which are potential sites for cross-linking. Since the crystal structure of OHase has not yet been solved, it is not known if these residues are exposed to the solvent or/and are close to the active site. However, given the fact that cross-linking is observed (vide infra) it is reasonable to assume that a certain percentage is located on the enzyme surface.

5.3.1 Selection of the best precipitating agent for CLEA preparation As a first step, several precipitating agents were screened, including organic solvents and an ammonium salt in an enzyme/precipitating agent ratio of 1:9 (v/v), both at 4°C and RT. To evaluate the different precipitating agents, the possibility to reactivate the aggregated enzyme after treatment (by dissolving in 20 mM Tris-HCl pH 8.0) is determined. As shown in Figure 5.2, the highest activity recovery was obtained when 2-propanol and ammonium sulfate were used. Using ammonium sulfate at RT results in more active enzyme, while no good activity recovery was observed for 2-propanol.

100 4°C 90 RT 80

70

60

50

40

30 Activity recovery [%] recovery Activity 20

10

0 Acetone Acetonitrile EtOH 2-Propanol DME (NH4)2SO4

Figure 5.2 Activity recovery of redissolved OHase aggregates after precipitation of 4 mg ml-1 CFE with 90% (v/v) of precipitation agent. Assays were performed using standard conditions in 20 mM Tris-HCl pH 8.0 with oleic acid as substrate. Enzyme activity of the free and soluble enzyme was designated as 100% activity.

This can be explained by the fact that hydrophilic solvents such as acetonitrile, ethanol or

116 5.3 Results and discussion acetone are able to take up infinite amounts of water, which in this case is stripped off from the enzyme surface [28]. As seen in Figure 5.2, this effect is for all solvents more pronounced at RT and is observed even for the less polar solvents such as 2-propanol and DME. For further studies both 2-propanol and ammonium sulfate procedures were used.

5.3.2 Cross-linking and the effect of glutaraldehyde concentration OHase CLEA’s were prepared from the precipitate with different concentrations of glutaraldehyde and three different incubation times, namely 1, 3 and 21 h at 4°C. In this step of CLEA preparation it is important to define proper reaction conditions in order to avoid excessive cross-linking, which can increase the rigidity of the enzyme and therefore negatively influence the performance of the enzyme. In contrast, using too low concentrations of the cross-linker may cause the formation of a highly flexible and small CLEA, which cannot be centrifuged and therefore is not suitable for reuse. From the results that are presented in Figure 5.3, a significant difference in activity recoveries was observed when ammonium sulfate (Figure 5.3A) and 2-propanol (Figure 5.3B) were used as precipitating agent. The overall activity recovery was 100-fold higher for ammonium sulfate than that for 2- propanol. The precipitation agent causes the enzyme to "freeze" in a certain conformation, which is subsequently covalently stabilized by glutaraldehyde. Despite the relatively high activity recovery obtained after precipitation of OHase with 2-propanol (Figure 5.2), hardly any activity recovery was observed once the precipitated OHase has been covalently cross- linked. This can be due to the fact that the precipitated OHase consisted of a rather unfavorable and inactive conformation, which could be reactivated after redissolving in buffer. On the contrary, this unfavorable conformation was stabilized after cross-linking with glutaraldehyde, thus resulting in inactive CLEA particles. Similar results were reported previously for a laccase from Trametes villosa, where good activity recoveries were obtained after precipitation with 2-propanol. However, once the enzyme was cross- linked, the catalytic activity decreased drastically [29].

117 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA)

(A) 16 1h 14 3h 12 21h 10 8 6 4 2 0 012345

(B) 0,15 Activity recovery [%] recovery Activity

0,10

0,05

0,00 012345 Glutaraldehyde [wt%] Figure 5.3 Preparation of OHase CLEA’s (4 mg ml-1) using ammonium sulfate (A) and 2-propanol (B) as precipitating agent in combination with different concentrations of the cross-linker glutaraldehyde. Cross- linking time used was 1, 3 and 21 h. CLEA activity recovery is determined by comparison with the total amount of units in 100 μl cell-free extract that was used for CLEA preparation.

Precipitation of OHase with ammonium sulfate resulted in significantly higher activity recoveries. The remaining activity in the CLEA was inversely proportional to the concentration of glutaraldehyde. With increasing glutaraldehyde concentration the quaternary protein structure probably deforms in such way that the enzyme loses its activity. Moreover, due to the high amount of lysine residues, the enzyme might become rigid and is not able to undergo conformational changes upon binding to the substrate. Therefore we chose a low glutaraldehyde concentration. Under these conditions a long cross-linking time gives better results. For further tests CLEA’s were prepared using 0.3wt% of glutaraldehyde and an incubation time of 21 h. No enzyme leakage was detected after 3 washing cycles with buffer when CLEA’s were prepared under these conditions. This indicates that despite the low glutaraldehyde concentration all enzyme molecules were properly cross-linked.

118 5.3 Results and discussion

Preparation of immobilized catalysts can introduce a new problem, i.e. diffusional limitations of the substrate and product due to the large catalyst particle size. Oleic acid is a C18 molecule that needs to be accommodated in the right position in the active site of OHase. However, if the catalyst forms large aggregates, the accessibility of the active site is impeded. This problem can be overcome by either reducing the aggregate particle size by mechanical stirring [30], or by reducing the amount of enzyme used in the CLEA preparation. We have prepared CLEA’s using 1 to 24 mg ml-1 of the CFE and the results are shown in Figure 5.4. The activity recovery was the highest for the smallest CFE concentration used. Although visually it was observed that more CLEA was obtained using 24 mg ml-1 of the CFE, it did not result in higher activity recovery. In general, the activity recovery decreases with increasing protein concentration. From all these experiments we deduced that the best way to make OHase CLEA’s is to prepare them with 90% ammonium sulfate, 0.3wt% glutaraldehyde and 1 mg protein ml-1 at 4°C using a cross-linking time of 21 h.

30

25

20

15

10 Activity recovery [%] recovery Activity

5

0 ABCD

Figure 5.4 Effect of enzyme concentration of the cell-free extract used for CLEA preparation on relative activity recovery of OHase CLEA. Cell-free extracts of 1 mg ml-1 (A), 4 mg ml-1 (B), 10 mg ml-1 (C) and 24 mg ml-1 (D) were used.

119 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA) 5.3.3 Thermal stability and pH profile of OHase CLEA’s The thermal stability of OHase as cell-suspension, in CFE and as CLEA was investigated by incubating the samples for 20 min at temperatures between 20 and 50°C in the absence of the substrate oleic acid and measuring their residual activity with the standard assay. The highest residual activity was obtained with CLEA (Figure 5.5). At 30°C the residual activity for the CLEA was significantly higher (81%) than that of the CFE (34%). When pre- incubated at 50°C, the CLEA still showed relative high activity (around 20%) compared to cell-suspension and CFE (5 and 3%, respectively). This phenomenon of increased temperature stability of CLEA has been observed earlier [30, 31]. This is probably due to enhancement of the structure stability through inter- and intramolecular covalent cross- links, which results in a molecule more resistant to conformational changes.

120 CLEA cell suspension cell-free extracts 100

80

60

40 Residual activity [%]

20

0 20 25 30 35 40 45 50 55 Temperature [°C]

Figure 5.5 Comparison of temperature stabilities of OHase as CLEA (□), in cell-free extract (○) and as cell- suspension (∆). Residual activities were assayed under standard conditions after the enzyme samples had been incubated at the indicated temperature (20 – 50°C) in 20 mM Tris-HCl pH 8.0 for 20 min. Initial activity determined under standard assay conditions was taken as 100%. Values are means of at least two independent measurements.

The pH dependency of OHase activity was investigated using oleic acid as substrate. From the results shown in Figure 5.6, it is obvious that OHase in CFE shows a different activity profile compared to CLEA or cell-suspension. The optimal pH for OHase as free soluble

120 5.3 Results and discussion

enzyme is at pH 7.0, on a broad plateau from pH 6.0 - 8.0, which is in agreement with previously reported results [11]. It is shifted to a smaller peak in more alkaline conditions (pH 8.0) for CLEA and cell-suspension. Acidic conditions (pH values 4.0 – 6.0) result in inactive CLEA and cell-suspension, while CFE shows about 57% of the relative activity at pH 5.0.

120

100

80

60

40 Residual activity [%]

20 CLEA cell suspension cell-free extracts 0 456789 pH

Figure 5.6 Comparison of pH profiles of OHase as CLEA (□), in cell-free extract (○) and as cell-suspension (∆). Relative activities were determined using oleic acid as substrate in buffers with varying pH (4.0 – 8.6) and compared to the activity measured using standard assay conditions (20 mM Tris-HCl pH 8.0).

There are many reasons, which could explain this different behavior. One important parameter in this study is the physical state of oleic acid in biological aqueous systems. Cistola and co-workers [32] discussed this in detail and concluded that oleic acid exists in three different states, which is dependent on the pH of the solution and on the ionic strength. At pH <7 oleic acid is in form of a stable oil phase and the carboxylic groups are protonated. From pH 7 - 9 the degree of ionization increases, which results in the formation of more structured lamellar system or large vesicles. Furthermore, this increase of ionization leads also to increased fluidity. If the concentration at pH >9 is above its critical micelle concentration (CMC ~6 μM), oleic acid starts to form micelles. Therefore, the pH dependency of enzymatic activities is not only the result of kinetic parameters but also

121 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA) depends on the nature of the substrates and the products. Firstly, the above mentioned fluidity of oleic acid at lower pH’s results in limited diffusivity, which again has big effect on OHase activity as cell-suspension or CLEA. Secondly, the apparent pKa of monomeric oleic acid is reported to be ~4.8, whereas it is 7.5 when incorporated into phospholipids bilayer of the cell membranes [32]. As CLEA is a densely packed hydrophobic environment, it may resemble a bilayer and have similar effects on the pKa of oleic acid. The charge variation of the substrate and the enzyme itself, and the resulting structural alterations may influence the binding of the substrate and therefore the catalytic activity of the enzyme.

5.3.4 Storage stability of OHase CLEA’s The storage stability of the OHase in CFE, as cell-suspension and as CLEA was determined at 4°C (Figure 5.7A) and RT (Figure 5.7B). In general, the residual activities for OHase in CFE and as CLEA were greater for samples stored at 4°C. Complete loss of activity was observed after 7 days of storage at RT. In contrast to that, the decrease of OHase activity over time as cell-suspension is significantly lower at RT than that at 4°C. While about 50% of the initial activity was lost after 7 days of storage at 4°C, the enzyme still retained 95% of its initial activity when stored at RT for the same period of time. The cross-linking of OHase leads to a slightly better storage stability at 4°C compared to the CFE. Nevertheless, after 3 days of storage the activity decreases at the same degree as OHase in CFE or as cell- suspension.

122 5.3 Results and discussion

(A) 120 CLEA cell suspension cell-free extract 100

80

60

40 Residual activity [%]

20

0 0246810121416 Time [days]

(B) 120 CLEA cell suspension 100 cell-free extract

80

60

40 Residual activity [%]

20

0 0246810121416 Time [days]

Figure 5.7 Comparison of storage stabilities at 4°C (A) and RT (B) of OHase as CLEA (□), in cell-free extract (○) and as cell-suspension (∆) in 20 mM Tris-HCl pH 8.0. Stabilities were given as residual activities, calculated by taking the initial activity of the enzyme as 100% under standard assay conditions.

123 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA) 5.3.5 Recycling of OHase CLEA’s For the recyclability studies, the CLEA’s were recovered and washed three times with buffer after each cycle of operation and the activity assay repeated with the same CLEA’s. The obtained results, which are depicted in Figure 5.8, show that the activity decreases with every cycle of operation. In order to eliminate the possible negative influence of HCl, which is needed for the extraction of fatty acids, we performed the extraction step without the addition of HCl and compared the results with those obtained for the standard procedure. Indeed, slightly higher residual activities were obtained when HCl was not used for the extraction, and were 2.7-fold higher after the second cycle and 6.5-fold higher after the third cycle. However, the loss of 70 and 87% of the initial activities after two and three cycles, respectively, clearly indicates low stabilities of the CLEA’s. This effect can be explained by the fact, that the product 10-HSA, which is a solid and starts to precipitate after a certain point of time, accumulates inside and outside the aggregated biocatalyst and blocks the structure for the diffusion of the substrate and the product. The inside portion will not be washed out under current washing conditions, resulting in new substrate not able to enter the CLEA’s and reach the active site of the OHase.

120 + HCl - HCl 100

80

60

40 Residual activity [%] activity Residual

20

0 123 Cycle number

Figure 5.8 Recycling stability of OHase CLEA. Standard activity assay was performed and after 3 x washing of the CLEA with buffer, the assay was repeated for a second and third time. The extraction step was performed with and without the addition of 3N HCl.

124 5.4 Conclusion

The same problem was observed by Cao et al., where penicillin G acylase CLEA starts to accumulate the product ampicilline inside the CLEA matrix [33]. Another immobilization technique may be considered at this point, such as soluble-insoluble supports [34]. With this technique it would be possible to retain the enzyme in solution while centrifuging off the solid product. With the subsequent lowering of the pH, the enzyme could be recovered in its insoluble form for the next cycle [34]. An additional option would be the preparation of OHase combined with another enzyme, such as lipase, which can convert 10-HSA to the soluble product lactone [35]. Preliminary results indicate that this procedure warrants further study.

5.3.6 Space-time yields Space-time yields for the production of 10-HSA by cell-suspension, CFE and CLEA’s were calculated and compared with previously reported values (Supplementary table 5.1). Cell- suspension and CFE produced roughly the same amount of the product 10-HSA with volumetric productivities of 0.26 and 0.33 g l-1 h-1, respectively. The space-time yields achieved with OHase CLEA’s were 4.7- and 6-fold lower than those obtained with the cell- suspension and CFE, respectively. Nonetheless, they were up to 55-fold higher than microbial productions with, for instance, Flavobacterium sp.DS5 (Supplementary table 5.1), performed under similar assay conditions (30°C) and with similar yield (~6%). Interestingly, in the listed studies (Supplementary table 5.1) all reported productivities above 5 g l-1 h-1 were achieved under optimized conditions, where surfactants or organic solvents were used to increase the solubility of the substrate and the product in the aqueous phase. It should be noted, however, that all obtained productivities in this study were performed under non-optimized conditions. Although, investigations of OHase immobilization as CLEA might not necessarily contribute directly to better 10-HSA productivities and certainly leave room for optimization, they might guide further investigations to make a stable and efficient biocatalyst for the industry.

5.4 Conclusion

This research describes the first steps towards a preparation of a biocatalyst for the production of 10-HSA. For the first time OHase from E. meningoseptica has been immobilized as cross-linked enzyme aggregates (CLEA’s). This immobilization technique

125 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA)

can be used to improve the biocatalytic properties of OHase. In the synthesis of 10-HSA, CLEA’s preparation of OHase led to a 2.4-fold increase of biocatalyst stability at elevated temperatures and better storage stabilities at cold temperatures in comparison with the soluble enzyme in cell-free extracts or as whole-cells. The perspectives that we give here could contribute to the preparation of a more successful biocatalyst for the application in this industrial process. [36-39]

126 5.5 Supplemental Information

5.5 Supplemental Information

Supplementary table 5.1 Comparison of space-time yield for the production of 10-HAS by OHase. ] 99 [36] 999 [36] 2012 this work 2012 this work 2009 [39] nd i 2012 this work asse 1997 [38] oritala 1989 [37] Author Year Ref. Kim STY [g/l/h] 0 16 Kim 2011 [18] Yield [%] Product [g/l] 0.9 0.8 89 0.011 Hudson 1995 [26] Substrate Substrate [g/l] 0.9 0.63 70 0.009 Hudson 1995 [26] ter ter 00 50 45.5 91 8.2 Jeon 2012 [15] 24h Assay Assay shaking [rpm] 24h Assay Assay pH Assay Assay temp. [°C] 35 6.5 NR 50 49 98 12.6 Joo 2012 [17] 35 7.5 35 200 6.5 NR 30 31.5 50 105 7.9 40 Kim 80 5 2010 [10] Joo 2012 [17] ic ic 40 6.5 25 150 6.8 180 17.8 18 16 8 90 2.7 K 44 1.6 Latr Assay Assay buffer Tween 40 80-anaerobic 80-anaerobic Tween 40 Growth Growth temp. [°C] Enrichment medium medium Enrichment 37 HB broth medium-anaerobic Enrichment 37 45/ 15 NR HB mediium af start 39 ~6.5 90 2 1.9 97 0.027 Morvan 19 medium medium TB-medium 28 0.02 M Tris-HCl 30 8 1400 1.7 0.26 15 0.26 Hiseni TB-medium 28 0.02 M Tris-HCl 30 8 1400 1.7 0.56 33 0.33 Hiseni TB-medium 28 0.02 M Tris-HCl 30 8 1400 1.7 0.09 5,3 0.055 Hisen mediuim 30 0.05 M K-phosphate 30 7.5 200 5.4 0.34 6 0.001 Heo a HB broth 45 HB mediium 39 ~6 90 2 1.52 76 0.021 Morvan 1999 [36 YMA NBY NR 30 0.05 M K-phosphate-anaerob K-phosphate-anaerob M 0.05 LB-medium LB-medium 16 EtOH % 4 + buffer PIPES M 0.05 35 6.5 NR 40 40 10 LB-medium LB-medium medium Riesenberg 16 25 % 0.05 + buffer Citrate-phosphate 80 Tween % 0.05 + 30 Tris-HCl M 0.05 7.5 2 HB broth 45/ 15 HB mediium 39 ~6.5 90 2 1.86 93 0.027 Morvan 1 Enrichment medium medium Enrichment 37 WF6Mn OA + 28 medium Growth medium-anaerobic Enrichment 37 28 NR NR Tween % 0.05 + Tris-HCl M 0.05 af start WF6Mn NR % 0.05 + buffer Citrate-phosphate 28 ~7 350 18 7 40 0.07 Kuo 2006 [8]

E. meningosepticaE.

sp.DS5 sp. gallinarum CFE containing Ohase Ohase containing CFE containing OHase containing Ohase containing Ohase containing Ohase Lysinibacillusfusiformis E. meningoseptica E. meningoseptica S. maltophilia S. maltophilia Enterococcus faecalis faecalis Enterococcus Enterococcus Flavobacterium E. coli E. from E. coli E. from (from recombinant recombinant (from Ohase) Lactobacillus Nocardia cholesterolicum N. Paraffinae E. coli from E. coli from P. acidilactici Selenomonas ruminantium Sphingobacterium thalpophilum Stenotrophmonas maltophilia coli E. from Stenotrophmonas nitritireducens NR: not reported reported not NR:

Strain Cultivation

CLEA

127 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA) 5.6 References

[1] R. A. Sheldon, "Green solvents for sustainable organic synthesis: state of the art," Green Chemistry, vol. 7, pp. 267-278, 2005. [2] N. Ran, et al., "Recent applications of biocatalysis in developing green chemistry for chemical synthesis at the industrial scale," Green Chemistry, vol. 10, pp. 361-372, 2008. [3] D. M. Alonso, et al., "Catalytic conversion of biomass to biofuels," Green Chemistry, vol. 12, pp. 1493-1513, 2011. [4] R. A. Sheldon, "Utilisation of biomass for sustainable fuels and chemicals: Molecules, methods and metrics," Catalysis Today, vol. 167, pp. 3-13, 2011. [5] A. S. Carlsson, "Plant oils as feedstock alternatives to petroleum - A short survey of potential oil crop platforms," Biochimie, vol. 91, pp. 665-670, Jun 2009. [6] S. H. Elsharkawy, et al., "Microbial oxidation of oleic-acid," Applied and Environmental Microbiology, vol. 58, pp. 2116-2122, Jul 1992. [7] L. L. Wallen, et al., "The microbiological production of 10-hydroxystearic acid from oleic acid," Archives of Biochemistry and Biophysics, vol. 99, pp. 249-253, 1962. [8] T. M. Kuo and W. E. Levinson, "Biocatalytic production of 10-hydroxystearic acid, 10-ketostearic acid, and their primary fatty amides," Journal of the American Oil Chemists Society, vol. 83, pp. 671-675, Aug 2006. [9] C. W. Seo, et al., "Hydration of squalene and oleic-acid by Corynebacterium sp. S-401," Agricultural and Biological Chemistry, vol. 45, pp. 2025-2030, 1981. [10] B. N. Kim, et al., "Conversion of oleic acid to 10-hydroxystearic acid by whole cells of Stenotrophomonas nitritireducens," Biotechnology Letters, vol. 33, pp. 993-997, May 2010. [11] L. E. Bevers, et al., "Oleate hydratase catalyzes the hydration of a nonactivated carbon-carbon bond," Journal of Bacteriology, vol. 191, pp. 5010-5012, Aug 2009. [12] E. N. Davis, et al., "Microbial hydration of cis-9-alkenoic acids " Lipids, vol. 4, pp. 356-362, 1969. [13] P. Marliere, "Method for producing an alkene comprising the step of converting an alcohol by an enzymatic dehydration step," 2011. [14] J. F. Jin and U. Hanefeld, "The selective addition of water to C=C bonds; enzymes are the best chemists," Chemical Communications, vol. 47, pp. 2502-2510, 2011. [15] E. Y. Jeon, et al., "Bioprocess engineering to produce 10-hydroxystearic acid from oleic acid by recombinant Escherichia coli expressing the oleate hydratase gene of Stenotrophomonas maltophilia," Process Biochemistry, vol. 47, pp. 941-947, 2012. [16] Y. C. Joo, et al., "Biochemical characterization and FAD-binding analysis of oleate hydratase from Macrococcus caseolyticus," Biochimie, vol. 94, pp. 907-915, 2012. [17] Y. C. Joo, et al., "Production of 10-hydroxystearic acid from oleic acid by whole cells of recombinant Escherichia coli containing oleate hydratase from Stenotrophomonas maltophilia," Journal of Biotechnology, vol. 158, pp. 17-23, 2012.

128 5.6 References

[18] B. N. Kim, et al., "Production of 10-hydroxystearic acid from oleic acid and olive oil hydrolyzate by an oleate hydratase from Lysinibacillus fusiformis," Applied Microbiology and Biotechnology, pp. 1-9, 2011. [19] C. Mateo, et al., "Improvement of enzyme activity, stability and selectivity via immobilization techniques," Enzyme and Microbial Technology, vol. 40, pp. 1451-1463, May 2007. [20] R. A. Sheldon, "Cross-linked enzyme aggregates (CLEAs): stable and recyclable biocatalysts," Biochemical Society Transactions, vol. 35, pp. 1583-1587, Dec 2007. [21] K. Hernandez and R. Fernandez-Lafuente, "Control of protein immobilization: Coupling immobilization and site-directed mutagenesis to improve biocatalyst or biosensor performance," Enzyme and Microbial Technology, vol. 48, pp. 107-122, Feb 2011. [22] R. A. Sheldon, "Cross-Linked Enzyme Aggregates as Industrial Biocatalysts," Organic Process Research & Development, vol. 15, pp. 213-223, 2011. [23] U. Hanefeld, et al., "Understanding enzyme immobilisation," Chemical Society Reviews, vol. 38, pp. 453-468, 2009. [24] C. Garcia-Galan, et al., "Potential of different enzyme immobilization strategies to improve enzyme performance," Advanced Synthesis and Catalysis, vol. 353, pp. 2885-2904, 2011. [25] R. A. Sheldon, "Enzyme immobilization: The quest for optimum performance," Advanced Synthesis & Catalysis, vol. 349, pp. 1289-1307, Jun 2007. [26] J. A. Hudson, et al., "Conversion of oleic acid to 10-hydroxystearic acid by two species of ruminal bacteria," Applied Microbiology and Biotechnology, vol. 44, pp. 1-6, Dec 1995. [27] A. Mehta, et al., "Rapid quantitation of free fatty acids in human plasma by high-performance liquid chromatography," Journal of Chromatography B, vol. 719, pp. 9-23, Nov 1998. [28] A. S. Bommarius and B. R. Riebel, "Biocatalysis in Non-conventional Media," in Biocatalysis, ed: Wiley-VCH Verlag GmbH & Co. KGaA, 2005, pp. 339-372. [29] I. Matijosyte, et al., "Preparation and use of cross-linked enzyme aggregates (CLEAs) of laccases," Journal of Molecular Catalysis B-Enzymatic, vol. 62, pp. 142-148, Feb 2010. [30] B. S. Aytar and U. Bakir, "Preparation of cross-linked tyrosinase aggregates," Process Biochemistry, vol. 43, pp. 125-131, Feb 2008. [31] M. E. Ortiz-Soto, et al., "Evaluation of cross-linked aggregates from purified Bacillus subtilis levansucrase mutants for transfructosylation reactions," Bmc Biotechnology, vol. 9, p. 68, 2009. [32] D. P. Cistola, et al., "Ionization and phase behavior of fatty acids in water: application of the Gibbs phase rule," Biochemistry, vol. 27, pp. 1881-1888, Mar 1988. [33] L. Q. Cao, et al., "Cross-linked enzyme aggregates: A simple and effective method for the immobilization of penicillin acylase," Organic Letters, vol. 2, pp. 1361-1364, May 2000. [34] J. Zhou, "Immobilization of cellulase on a reversibly soluble-insoluble support: Properties and application," Journal of Agricultural and Food Chemistry, vol. 58, pp. 6741-6746, 2010. [35] A. Hiseni, et al., "Biochemical characterization of the carotenoid 1,2-hydratases (CrtC) from Rubrivivax gelatinosus and Thiocapsa roseopersicina," Applied Microbiology and Biotechnology, vol. 91, pp. 1029-1036, 2011.

129 Preparation and properties of immobilized oleate hydratase as a cross-linked enzyme aggregate (CLEA)

[36] B. Morvan and K. N. Joblin, "Hydration of oleic acid by Enterococcus gallinarum, Pediococcus acidilactici and Lactobacillus sp. isolated from the rumen," Anaerobe, vol. 5, pp. 605-611, 1999. [37] S. Koritala, et al., "Microbial conversion of oleic acid to 10-hydroxystearic acid," Applied Microbiology and Biotechnology, vol. 32, pp. 299-304, 1989. [38] A. Latrasse, et al., "Conversion of oleic acid to 10-hydroxystearic acid by Nocardia paraffinae," Biotechnology Letters, vol. 19, pp. 715-718, 1997. [39] S. H. Heo, et al., "Production of oxygenated fatty acids from vegetable oils by Flavobacterium sp. strain DS5," New Biotechnology, vol. 26, pp. 105-108, 2009.

130

Chapter 6

6 Conclusions and future prospects

Conclusions and future prospects

The study described in this thesis was set out to explore the potential of newly discovered hydratases, carotenoid 1,2-hydratases (CrtC) from photosynthetic bacteria Rubrivivax gelatinousus and Thiocapsa roseopersicina and oleate hydratase (OHase) from Elizabethkingia meningoseptica, for their use as biocatalysts in industrial processes. In order to do so, it was important to gain more insight in the activity, stability substrate scope and structure-function relationship of these enzymes. When the research described in this thesis started, limited or no literature was available on structural and mechanistic properties of these two groups of hydratases. The work presented in this thesis sought to broaden the structural, mechanistic and sequence-based knowledge required to define future research. The main findings of this thesis are presented in the ‘Summary’ section of this thesis. This chapter will evaluate the original question, which was posed at the start of the research project, namely to evaluate if carotenoid 1,2-hydratases and oleate hydratases have potential as industrial biocatalysts.

6.1 Carotenoid 1,2-hydratase

A comparative study of homologous carotenoid 1,2-hydratases sequences was successfully applied to identify conserved residues within selected regions across the gene. Next to mutagenesis of selected key residues and biochemical methods, which have proven to be effective for determining the potential catalytic mechanism, the elucidation of a three- dimensional structure is also pivotal to unravel the structure-function relationship of this group of enzymes. Although, we made a homology model, which appeared to be accurate since it can explain the mutants we made, the availability of a three-dimensional structure would allow the application of site-saturated mutagenesis in residues governing the property of interest and the generation of synergistic effects of neighboring mutations. In addition, docking models can be used to predict the preferred orientation of substrates and residues involved. The work presented in this thesis has helped to shed light on structural and mechanistic properties of carotenoid 1,2-hydratases. But also, it has clearly shown that their low activities at present might prevent their successful utilization in industrial processes. The fact that they are membrane-bound may have some advantages as it can be regarded as natural immobilization. On the other hand, most enzymes used in industrial settings are extracellular enzymes, as they require less downstream processing prior to catalysis and

132 Carotenoid 1,2-hydratase

they are more stable to external environmental perturbations. The overproduction of carotenoid 1,2-hydratases is primarily limited to the membrane capacity of the host organism. Therefore, the development of an effective expression system is of a high importance. For instance, the usage of a host that is less sensitive to the toxic effects of overexpressed membrane or membrane-bound proteins, which have possibly formed aggregates due to space constrains, will form more biomass that can produce more protein [1-6]. Another limitation of carotenoid 1,2-hydratases is the very low specific activity. This might be due to either non-optimal assay conditions or the intrinsic property of membrane-bound enzymes. High specific activity ensures sufficient space-time yield and makes the process and reactor costs low. Therefore, the catalytic activity of these hydratases needs to be improved, either by random mutagenesis, followed by site-saturated mutagenesis as a means to further improve beneficial mutations obtained through random mutagenesis, or by immobilization. As demonstrated by Umeno et al. [7] the evolvability of carotenoid biosynthetic enzymes is remarkable. Carotenoid biosynthesis pathway appears to consist of two groups of enzymes: ‘gatekeeper’ enzymes, which are located at the earliest steps of the pathway and/or at important branch points and enzymes that are ‘locally specific’, i.e. they recognize a particular structural motif of a possible substrate. The former dictate the product diversity by allowing only certain molecules to enter the pathway and are therefore expected to be very specific. Several researches have demonstrated the ability of these enzymes to acquire new specificities after a limited genetic change. In contrast to the high specificity of the ‘gatekeeper’ enzymes, ‘locally specific’ enzyme do not require this property as they are placed at several locations in the pathway and are supposed to convert more than one substrate. In case of a change upstream, which would allow a new compound to enter the pathway, downstream enzymes are expected to be able to also metabolize this new compound. Carotenoid 1,2-hydratase is such an enzyme that is less specific and appears to recognize only a particular motif of the structure (-end group according to carotenoid nomenclature). Thus, this fact predicts a good prospect for this enzyme to metabolize unnatural substrates as long as it retains the required motif. Overall, if it would be possible to address these two main issues, i.e. overexpression and catalytic activity, this would significantly enlarge the potential of carotenoid 1,2- hydratases. Future research could be the focus on optimizing additional enzyme properties, such as substrate promiscuity, thermostability, solvent tolerance and activity at high

133 Conclusions and future prospects

substrate concentration. Nevertheless, the time that needs to be invested in order to make this enzyme suitable for industrial process has to be balanced against the chances for success in reasonable timeframe.

6.2 Oleate hydratase

Another hydratase with high potential for industrial biocatalysis is oleate hydratase. Since the first paper was published on characterization of recombinantly overexpressed oleate hydratase by Hagen et. al [8], the literature on oleate hydratase significantly expanded with several papers, mostly with the work from the group of Oh [9-15] and Feussner [16, 17]. While the production of 10-hydroxystearic acid in high yields and biochemical characterization of several recombinantly overexpressed oleate hydratases have been extensively studied by Oh et al., the latter author succeeded very recently in elucidating the first three-dimensional structure of the oleate hydratase from Lactobacillus acidophilus (LAH). Despite the significant contribution of this work to the better understanding of structural properties of LAH, the precise location of the active site and the involved catalytic mechanism still remain unclear. The researchers were able to elucidate the crystal structure of an apo-LAH and LAH co-crystallized with linoleic acid (LA-LHA), but both without the co-factor FAD (lost during enzyme purification). In addition, the binding of the substrate LA in the active site, was hindered by two MPD (2-methyl-2,4-pentanediol; used as precipitating agent) molecules so that it stayed bound at the entrance of the substrate channel. Nevertheless, given the fact that the LAH shares more than 50% amino acid sequence similarity with the oleate hydratase from E. meningoseptica, which was studied in this thesis work, it is likely to assume that these novel structural insights will partly help to increase our understanding of the structural properties of oleate hydratases. Future research of the study presented in this thesis, however, could attempt to crystallize oleate hydratase from E. meningoseptica in the presence of a substrate analog in order to locate the active site and amino acids necessary for binding of the substrate and catalysis. Once the catalytic mechanism is unraveled and together with other newly gained structural data, the analysis of the created mutants (Chapter 4) can be performed in more detail, which again could give an indication of directions for further improvement of the catalytic activity or substrate specificity.

134 High-throughput screening assay

With regard to formulating the enzyme by means of immobilization through cross-linking (Chapter 5) in order to make it suitable for an economically feasible industrial process, other methods than cross-linking need to be explored. As the product 10-hydroxystearic acid is a solid, a method needs to be chosen which prevents the accumulation of the solid product within the immobilized enzyme complex. This might be accomplished by immediate removal of the product either by a physical isolation through a second phase or by reacting the product further to produce a water soluble intermediate or a final product. Overall, the good activity of OHase together with its straightforward expression system makes this enzyme a promising candidate for future directed evolution studies.

6.3 High-throughput screening assay

A crucial part of any directed evolution study is a robust and selective high-throughput screening assay, which can be used to isolate mutants with improved properties. The assay allows testing of large number of mutants in a quick and efficient way. The developed colorimetric high-throughput screening assay (Figure 6.1) with oleate hydratase as model enzyme (Chapter 4) that can be used for the detection of primary, secondary and tertiary alcohols, provides an adequate basis for screening a large amount of generated mutants. Although the assay has been proven to be robust and applicable for many different substrate/product pairs, the signal window needs further improvement. The current status is that by using purified oleate hydratase instead of crude E. coli extracts, the signal window could significantly be increased, allowing sufficient discrimination between the noise and enzymatic activity. This result increases the chance of finding mutants with improved properties, without having the risk of detecting false positive mutants. Although enzyme purification can be performed in a high-throughput manner, it would be of more interest to follow optimization approaches that prevent too high signal of the noise or that increase the overexpression yield of the enzyme in a microtiter plate. Besides the colorimetric assay developed by us, also calorimetric assays could contribute to HTS of activity of these enzymes [18]. Calorimetry measures the absorbed or exerted heat during bond breaking or making. This means that this assay is the universal assay for enzymatic activity using real substrates instead of artificial ones. It is already shown in our lab that the enthalpy of hydration is large enough to measure enzyme kinetics. Overall, this

135 Conclusions and future prospects will allow the rate evaluation of activity, stability and substrate scope of large libraries of mutants.

ACTIVITY SOURCE

DATA PROCESSING ACTIVITY/SCREENING

ASSAY

ABSORPTION

MEASUREMENT

PLATE PROCESSING

Figure 6.1 Schematic overview of the high high-throughput screening assay. The assay was developed for the detection of enzymatic hydration activity assessed by selective spectrophotometric detection of alcohols.

Next to OHase and CrtC, other hydratases that also operate on (non-activated) double bonds are continuously discovered by microbiologists [19]. Glycerol dehydratases (EC 4.2.1.30) might be of interest for the dehydration of glycerol (a huge by-product of biodiesel production) to get the building block 3-hydroxypropanol [20]. Also the very fast enzyme enoyl-CoA hydratase (EC 4.2.1.17) could be investigated for promiscuous activities [21]. Although the promiscuous activity of an enzyme is usually orders of magnitudes lower than the activity with the natural substrate, this could still be acceptable because of the almost diffusion limited rate of this enzyme. The linalool dehydratase-isomerase (EC 4.2.1.127) might be an interesting enzyme when looking at fatty acids, since it has its function in the production of myrcene from linalool, a C10-terpene [22]. Hydroxycinnamoyl-CoA hydratase-lyase (EC 4.2.1.101) could be used in the biosynthesis of the flavour compound vanillin [23]. A key step is the addition of water to the thioester of ferulic acid, catalysed by hydroxycinnamoyl-CoA hydratase lyase (HCHL, formerly feruloyl-CoA hydratase). The hydration is immediately followed by a retro-aldol reaction, releasing the vanillin and acetyl-CoA [24].

136 High-throughput screening assay

In conclusion, I believe there is a bright future for hydratase enzymes and that there are enough leads for follow-up research on the enzymatic hydration of double bonds. Our assay can serve to explore these novel enzymes for their potential.

137 Conclusions and future prospects

6.4 References

[1] S. Wagner, et al., "Tuning Escherichia coli for membrane protein overexpression," Proceedings of the National Academy of Sciences, vol. 105, pp. 14371-14376, September 23, 2008 2008. [2] G. J. Gopal and A. Kumar, "Strategies for the production of recombinant protein in escherichia coli," Protein Journal, vol. 32, pp. 419-425, 2013. [3] N. Gul, et al., "Evolved escherichia coli strains for amplified, functional expression of membrane proteins," Journal of Molecular Biology, vol. 426, pp. 136-149, 2014. [4] T. M. Lo, et al., "Microbial engineering strategies to improve cell viability for biochemical production," Biotechnology Advances, vol. 31, pp. 903-914, 2013. [5] D. M. Molina, et al., "Engineering membrane protein overproduction in Escherichia coli," Protein Science, vol. 17, pp. 673-680, 2008. [6] E. Gordon, et al., "Effective high-throughput overproduction of membrane proteins in Escherichia coli," Protein Expression and Purification, vol. 62, pp. 1-8, 2008. [7] D. Umeno, et al., "Diversifying carotenoid biosynthetic pathways by directed evolution," Microbiology and Molecular Biology Reviews, vol. 69, pp. 51-78, Mar 2005. [8] L. E. Bevers, et al., "Oleate hydratase catalyzes the hydration of a nonactivated carbon-carbon bond," Journal of Bacteriology, vol. 191, pp. 5010-5012, Aug 2009. [9] B. N. Kim, et al., "Production of 10-hydroxystearic acid from oleic acid and olive oil hydrolyzate by an oleate hydratase from Lysinibacillus fusiformis," Applied Microbiology and Biotechnology, pp. 1-9, 2011. [10] B. N. Kim, et al., "Production of 10-hydroxystearic acid from oleic acid and olive oil hydrolyzate by an oleate hydratase from Lysinibacillus fusiformis," Applied Microbiology and Biotechnology, vol. 95, pp. 929-937, 2012. [11] B. N. Kim, et al., "Conversion of oleic acid to 10-hydroxystearic acid by whole cells of Stenotrophomonas nitritireducens," Biotechnology Letters, vol. 33, pp. 993-997, May 2010. [12] J. S. Kim, et al., "Identification and characterization of a novel nitrilase from Pseudomonas fluorescens Pf-5," Applied Microbiology and Biotechnology, vol. 83, pp. 273-283, May 2009. [13] K. R. Kim and D. K. Oh, "Production of hydroxy fatty acids by microbial fatty acid-hydroxylation enzymes," Biotechnology Advances, 2013. [14] Y. C. Joo, et al., "Biochemical characterization and FAD-binding analysis of oleate hydratase from Macrococcus caseolyticus," Biochimie, vol. 94, pp. 907-915, 2012. [15] Y. C. Joo, et al., "Production of 10-hydroxystearic acid from oleic acid by whole cells of recombinant Escherichia coli containing oleate hydratase from Stenotrophomonas maltophilia," Journal of Biotechnology, vol. 158, pp. 17-23, 2012. [16] A. Volkov, et al., "Crystal structure analysis of a fatty acid double-bond hydratase from Lactobacillus acidophilus," Acta Crystallographica Section D: Biological Crystallography, vol. 69, pp. 648-657, 2013.

138 References

[17] A. Volkov, et al., "Myosin cross-reactive antigen of Streptococcus pyogenes M49 encodes a fatty acid double bond hydratase that plays a role in oleic acid detoxification and bacterial virulence," Journal of Biological Chemistry, vol. 285, pp. 10353-10361, 2010. [18] M. J. Todd and J. Gomez, "Enzyme kinetics determined using calorimetry: A general assay for enzyme activity?," Analytical Biochemistry, vol. 296, pp. 179-187, 2001. [19] J. F. Jin and U. Hanefeld, "The selective addition of water to C=C bonds; enzymes are the best chemists," Chemical Communications, vol. 47, pp. 2502-2510, 2011. [20] S. Kwak, et al., "Biosynthesis of 3-hydroxypropionic acid from glycerol in recombinant Escherichia coli expressing Lactobacillus brevis dhaB and dhaR gene clusters and E. coli K-12 aldH," Bioresource Technology, vol. 135, pp. 432-439, 2013. [21] G. Agnihotri and H. W. Liu, "Enoyl-CoA hydratase: Reaction, mechanism, and inhibition," Bioorganic and Medicinal Chemistry, vol. 11, pp. 9-20, 2003. [22] D. Brodkorb, et al., "Linalool dehydratase-isomerase, a bifunctional enzyme in the anaerobic degradation of monoterpenes," Journal of Biological Chemistry, vol. 285, pp. 30436-30442, 2010. [23] P. M. Leonard, et al., "The 1.8 Å resolution structure of hydroxycinnamoyl-coenzyme a hydratase- lyase (HCHL) from Pseudomonas fluorescens, an enzyme that catalyses the transformation of feruloyl-coenzyme A to vanillin," Acta Crystallographica Section D: Biological Crystallography, vol. 62, pp. 1494-1501, 2006. [24] J. P. Bennett, et al., "A ternary complex of hydroxycinnamoyl-CoA hydratase-lyase (HCHL) with acetyl-CoA and vanillin gives insights into substrate specificity and mechanism," Biochemical Journal, vol. 414, pp. 281-289, 2008.

139

Summary/Samenvatting

Summary/Samenvatting

Summary

The rapid development in the field of biotechnology over the last four decades, in addition to an increasing recognition that we have limited resources and thus need to move to renewable raw materials, have been drivers for the chemical industry to look at enzymes as novel catalysts. In addition, enzymes are highly specific, thereby leading to high regio- and chiral selectivities and less/no unwanted side reactions and byproducts. They generally operate under mild conditions, resulting in energy savings. Overall, it is safe to state that enzymes contribute to the environmentally sustainable processing. Hydratases catalyze the non-hydrolytic and non-oxidative addition and/or removal of a water molecule to a carbon-carbon double bond. From a chemical point of view, this reaction is difficult to achieve and requires harsh conditions, such as high temperature and low pH. In contrast, the enzymatic route proceeds under very mild conditions in a neutral aqueous environment, yielding products in high yields and without undesired side reactions. Therefore, there is significant interest in the application of hydratases as efficient, selective and environmentally friendly biocatalyst. The research of this thesis focused on two hydratases: carotenoid 1,2-hydratase (CrtC) and oleate hydratase (OHase). CrtC is an enzyme found in the biosynthetic pathway of carotenoids. CrtC introduces a tertiary hydroxyl group into a carotenoid molecule by addition of water to the carbon- carbon double bond at the C-1 position. Another hydratase that has raised the attention of researchers is OHase. OHase catalyzes the conversion of oleic acid (OA) into (R)-10- hydroxystearic acid (10-HSA), a high-added value product used for the production of a number of products, including resins, nylons, plastics, waxes, cosmetics and coatings. This hydratase, as well as the carotenoid 1,2-hydratase, represents a new type of hydro-lyase as it is able to hydrate an isolated carbon-carbon double bond. In literature, a limited amount of data was available on the biochemical, structural and mechanistic properties of these two hydratases. Therefore, it was decided to study these enzymes with a focus on their structure-function relationship, thus allowing the evaluation of the potential of these hydratases as industrial biocatalysts. In Chapter 1, a general overview is given on enzymes and their application as biocatalysts in various industries. Also, protein engineering tools used to overcome the limitations of natural enzymes as biocatalysts at typical operating industrial conditions, such as high substrate and salt concentrations, use of organic solvents, etc., are introduced. Our present

142 Summary/Samenvatting

knowledge on hydro-lyases and their utilization in industrial processes is highlighted. Special attention is given to aspects of the structure-function relationship of the two studied hydratases CrtC and OHase. Chapter 2 describes the detailed biochemical characterization of two newly discovered CrtC’s from photosynthetic bacteria Rubrivivax gelatinosus and Thiocapsa roseopersicina. In order to investigate the biochemical properties, the enzymes were recombinantly overexpressed and purified by affinity chromatography. It was demonstrated that both CrtC’s were able to cofactor independently catalyze the conversion of the natural substrate

lycopene to 1-HO- and 1,1′-(HO)2-lycopene. In addition, low activity was detected with an unnatural substrate geranylgeraniol (C20 substrate), which functionally resembles the natural C40 substrate lycopene. Both CrtC’s are stable at a broad and suitable temperature and pH range, which makes them attractive for green hydration reactions in industrial applications. Although, the amino acid sequences of RgCrtC and TrCrtC differ by only one amino acid (406 vs. 405), a structural difference has been observed by means of SDS- PAGE and MS analysis. Whereas RgCrtC is expressed as a 44 kDa protein, TrCrtC exist as a 38 kDa protein, most likely caused by autocatalytic processing. In order to increase our understanding of the structure and mechanism of CrtC’s from photosynthetic bacteria, protein engineering techniques site-directed evolution and semi- rational mutagenesis were applied (Chapter 3). By generating alanine point-mutants of selected amino acid positions, it was possible to elucidate the role of the amino acids His239, Trp241, Tyr266 and Asp268 in RgCrtC (and the corresponding amino acids in TrCrtC) and identify them as key residues, which are directly involved in the catalytic reaction. By analyzing a partial 3D structure obtained by homology modeling with the distantly related putative AttH protein from Nitrosomonas europaea it could be shown that all identified amino acids were in close proximity to each other. All these results indicate that the aforementioned amino acid residues are involved in the catalytic cycle. When considering the generation of tailor-made biocatalysts for a successful utilization in industrial processes, the availability of a suitable high-throughput screening or selection method is a pre-requisite. The existence of such an assay will also determine the method of choice for protein engineering. Due to limited existing structural and mechanistic knowledge on hydratases studied in this thesis, a high-throughput screening assay for the detection of alcohols, products of hydrating enzymes such as CrtC and OHase, was developed (Chapter 4), which allows rapid screening of a large number of variants within a reasonable timeframe. OHase from Elizabethkingia meningoseptica was used as the

143 Summary/Samenvatting

model enzyme to examine and characterize the capability of the developed method for enabling an automated set up. The assay was able to detect primary, secondary and tertiary alcohols in the presence of fatty acids as well as small cyclic and acyclic unsaturated alkenes as substrates. Besides protein engineering techniques to improve the operational performance of enzymes (e.g. thermo-stability, activity and solvent tolerance) immobilization is a convenient approach towards stabilization. In Chapter 5 we report on the immobilization of OHase as cross-linked enzyme aggregates (CLEA). For this purpose, recombinant OHase from E. coli cell-free extracts was aggregated and cross-linked using a bifunctional cross-linker glutaraldehyde. With an activity recovery of 26% after 21h cross- linking at 4°C, CLEA’s preparation of OHase led to a 2.4-fold increase of biocatalyst stability at elevated temperatures and better storage stabilities at cold temperatures. Furthermore, up to 55-fold higher space-time yields were achieved with OHase CLEA’s compared to microbial productions. Overall, the work presented in this thesis has contributed to an understanding of the structure-function relationship of two newly discovered hydratases: carotenoid 1,2- hydratase and oleate hydratase. Furthermore, it may contribute to the development of a biocatalyst that can be used for the production of high-added value compounds in industrial processes.

144 Summary/Samenvatting

Samenvatting

De snelle ontwikkelingen op het gebied van de biotechnologie in de afgelopen vier decennia, samen met de toegenomen bewustwording dat we over beperkte hoeveelheden aan fossielen grondstoffen beschikken, en er een noodzaak bestaat om over te gaan naar hernieuwbare grondstoffen, zijn aanleiding geweest voor de chemische industrie om enzymen te ontwikkelen als nieuwe en hernieuwbare katalysatoren. De selectiviteit van enzymen zorgt ervoor dat minder of geen ongewenste nevenreacties en bijproducten worden geproduceerd. Daarnaast werken enzymen onder milde condities, wat bijdraagt aan energiezuinige procescondities. Over het algemeen kan men dus stellen dat het gebruik van enzymen als katalysatoren kan bijdragen aan duurzame processen. Hydratases katalyseren de niet-hydrolytische en niet-oxidatieve additie en/of eliminatie van een watermolecuul aan een koolstof-koolstof dubbele binding. Vanuit chemisch oogpunt is deze reactie moeilijk uit te voeren en vereist extreme condities, zoals hoge temperaturen en lage pH. Daarentegen verloopt de enzymatische route onder zeer milde omstandigheden in een neutrale waterige omgeving, waardoor producten in hoge opbrengsten en zonder ongewenste nevenreacties gemaakt kunnen worden. Daarom is het zeer relevant om hydratases te bestuderen als efficiënte, selectieve en milieuvriendelijke biokatalysatoren. Het onderzoek van dit proefschrift richt zich op twee enzymen: carotenoïd 1,2- hydratase (CrtC) en oleaat hydratase (OHase). CrtC is een enzym dat aanwezig is in de biosynthese route van carotenoïden. CrtC introduceert een tertiaire hydroxyl groep in een carotenoïde molecuul door toevoeging van water aan de koolstof-koolstof dubbele binding op de C-1 positie. Een ander hydratase dat de aandacht van onderzoekers heeft getrokken is OHase. OHase katalyseert de omzetting van oliezuur (OA) in (R)-10-hydroxystearinezuur (10-HSA), een product met hoge toegevoegde waarde voor de productie van materialen zoals harsen, nylon, kunststoffen, wassen, cosmetica en coating. Dit hydratase, evenals de carotenoïd 1,2-hydratase, vertegenwoordigt een nieuw type hydro-lyase dat in staat is om een geïsoleerde koolstof- koolstof dubbele binding te hydrateren. In de literatuur is een beperkte hoeveelheid gegevens beschikbaar met betrekking tot de biochemische, structurele en mechanistische eigenschappen van deze twee hydratases. Daarom werd besloten om deze enzymen te bestuderen om meer inzicht te krijgen in de

145 Summary/Samenvatting structuur-functie relatie en zodoende het potentieel van hydratases als biokatalysatoren in industriële processen in kaart te brengen. In Hoofdstuk 1 wordt een algemeen overzicht gegeven van enzymen en hun toepassing als biokatalysatoren in diverse industrieën. Ook ‘protein engineering’ technieken die worden gebruikt om de beperkingen van de natuurlijke enzymen als biokatalysatoren bij typische operationele industriële omstandigheden te overwinnen, zoals hoge substraat- en zoutconcentraties, gebruik van organische oplosmiddelen, etc., worden geïntroduceerd. Onze huidige kennis over hydro-lyasen en hun gebruik in industriële processen wordt daarbij benadrukt. Speciale aandacht wordt besteed aan aspecten van de structuur-functie relatie van de twee bestudeerde hydratases CrtC en OHase. Hoofdstuk 2 beschrijft de gedetailleerde biochemische karakterisering van twee nieuw ontdekte CrtC’s van de fotosynthetische bacteriën Rubrivivax gelatinosus en Thiocapsa roseopersicina. Om de biochemische eigenschappen te onderzoeken, werden de enzymen recombinant tot overexpressie gebracht en gezuiverd middels affiniteitschromatografie. Er werd aangetoond dat beide CrtC’s zonder de hulp van een cofactor de omzetting van het natuurlijke substraat lycopeen naar 1-HO- en 1,1'-(HO)2-lycopeen konden katalyseren. Bovendien werd lage activiteit gedetecteerd met het niet-natuurlijke substraat geranylgeraniol (C20 substraat), dat structureel lijkt op het natuurlijke substraat lycopeen. Beide CrtC’s zijn stabiel in een breed en gemiddeld temperatuur- en pH-bereik, waardoor ze aantrekkelijk worden voor groene hydratatie reacties in industriële toepassingen. Hoewel de theoretische eiwitgrootte van RgCrtC en TrCrtC slechts in één aminozuur verschilt (406 versus 405) is een structureel verschil waargenomen door middel van SDS-PAGE en MS- analyse. Terwijl RgCrtC als een 44 kDa eiwit tot expressie wordt gebracht, bestaat TrCrtC als een 38 kDa eiwit, waarschijnlijk veroorzaakt door autokatalytische verwerking. Om onze kennis van de structuur en het mechanisme van CrtC’s uit fotosynthetische bacteriën te verhogen, werden protein engineering technieken semi-gerichte evolutie (in het engels semi-directed evolution) en semi-rationale mutagenese toegepast (Hoofdstuk 3). Door het genereren van specifieke alanine punt-mutanten van geselecteerde aminozuurposities, was het mogelijk om de rol van de aminozuren His239, Trp241, Tyr266 en Asp268 in RgCrtC (en de overeenkomstige aminozuren in TrCrtC) te verduidelijken en te identificeren als belangrijke residuen die direct betrokken zijn bij de katalytische reactie. Door het analyseren van een gedeelte van de 3D-structuur, die verkregen is door homologie modellering met het verwante AttH eiwit van Nitrosomonas europaea, kon worden aangetoond dat alle geïdentificeerde aminozuren zich in directe omgeving van elkaar

146 Summary/Samenvatting

bevinden. Al deze resultaten zijn een eerste aanwijzing dat deze aminozuren betrokken zijn bij de katalytische cyclus. Voor onderzoek naar ‘op maat gemaakte biokatalysatoren’ is de beschikbaarheid van een ‘high-throughput screening’ of selectie methode een eerste vereiste. Het bestaan van een dergelijke methode zal ook de keuze van de ‘protein engineering’ methode bepalen. Een high-throughput screening test is ontwikkeld voor de detectie van alcoholen, de producten van hydraterende enzymen zoals CrtC en OHase (Hoofdstuk 4). OHase van Elizabethkingia meningoseptica werd als model enzym gebruikt om het vermogen van de ontwikkelde methode voor het mogelijk maken van een geautomatiseerde opstelling te onderzoeken en te karakteriseren. De test bleek in staat om primaire, secundaire en tertiaire alcoholen te detecteren in de aanwezigheid van de start-verbinding: onverzadigde vetzuren en kleine cyclische en niet-cyclische onverzadigde alkenen als substraten. Naast ‘protein engineering’ technieken, die worden gebruikt om operationele prestaties van enzymen (bijvoorbeeld thermostabiliteit, activiteit en oplosmiddel tolerantie) te verbeteren, wordt ook immobilisatie toegepast voor stabilisatie van enzymen. In Hoofdstuk 5 beschrijven we de immobilisatie van OHase als verknoopte enzym-aggregaten (in het Engels cross-linked enzyme aggregates (CLEA)). Hiervoor wordt het recombinante OHase uit E. coli celvrije extracten geaggregeerd en verknoopt met een bi-functionele crosslinker glutaaraldehyde. Met een activiteitsbehoud van 26% na 21 uur verknoping bij 4°C, leidde de CLEA bereiding van OHase tot een 2,4-voudige toename van biokatalysator stabiliteit bij verhoogde temperaturen en een betere opslagstabiliteit bij lage temperaturen. Bovendien werden tot 55-voudig hogere ruimte-tijd rendement (in het Engels space-time yield) bereikt met OHase CLEA’s ten opzichte van microbiële productie. Samenvattend heeft het werk dat in dit proefschrift is uitgevoerd, bijgedragen aan een begrip van de structuur-functie relatie van twee pas ontdekte hydratases: carotenoïd 1,2- hydratase en oleaat hydratase. Bovendien kan het werk bijdragen tot de ontwikkeling van een biokatalysator die kan worden gebruikt voor de productie van stoffen met hoge toegevoegde waarde in industriële processen.

147

Acknowledgements

Acknowledgements The completion of this thesis has been a long journey, but I did it! At this point I would like to thank all the important people who have contributed to this thesis in one way or another. First, I am grateful to my promotor Prof. Isabel Arends for her guidance and encouragement throughout these years. You are truly an inspiration for me. Then, my deepest gratitude goes to my daily supervisor Dr. Linda Otten. Linda, your absolutely supportive, positive attitude towards all aspects of my research was a great help. Certainly, we had to find our way in the beginning as I was your first PhD student, but we succeeded in finding a good way that worked for us (for sure, the fact that we were sitting in the same office helped a lot). It was your support and encouragement, which helped me to also manage all the difficult phases of my PhD time. You have been through a very difficult phase yourself, but even then, you managed to always find time when I needed advice, inspiration or critical comments. I am very glad that I have met you and I will always be grateful for everything that you have done for me. You are a great person. I would also like to thank Martin Gorseling for all the technical support. Martin, I am glad that you have found the way to our group. You have made a very big, positive change in our group and I am for sure not the only one who appreciates everything that you have done. No matter what technical problem I had, you managed always to solve it very quickly. Your knowledge about all apparatus is amazing and I learned a lot from you. Now, I would like to take this opportunity to thank Mieke van der Kooij. Thank you for all your help in administrative matters and beyond. It is amazing how you keep track on everything and never forget to send a reminder if something is about to expire. Although, we did not have many opportunities to talk, I always enjoyed our few short conversations in your office. Special thanks go to my paranymph Rosario Franco Berriel. You are an amazing person and working with you was a lot of fun. I always enjoyed our lunches together and our conversations about ‘Gott und die Welt’. Thank you for you friendship and I wish you and your cute family the best for your future. I extend my gratitude to the ‘other’ Rosario, Rosario Medici. I am amazed about your knowledge and professionalism. Working hard and focused, helping each other and performing experiments carefully and precisely, these all comes to my mind when I think about you. I appreciate all the work that you have done in order to make the HTS-assay

149 Acknowledgements publishable. I truly enjoyed working together with you and being friend with you. All the best for your professional carrier and your lovely family. Then, my colleagues and officemates from ENZ and BOC, whose longer or shorter presence enriched the life at work. Thank you all for the nice time during my stay at the TU-Delft and all the support. I was privileged to work in two groups and to learn from all of you. All the supports from my family and my family in-low are highly appreciated. I am indebted to them. Finally, my greatest thanks goes to my friend, my soul mate, my beloved husband Senad. Thank you for all your love, support and understanding.

150 Curriculum vitae

Curriculum vitae Aida Hiseni was born on May 20, 1980, in Doboj (Bosnia and Hercegovina). She pursued studies in Biology at the Heinrich-Heine-Universität, Düsseldorf, Germany, where she received the Diplom degree in Biology in 2007. In the same year she moved to the Netherlands and commenced her Ph.D. work in Biotechnology at the Delft University of Technology. Since November 2011, Aida works as associate scientist at DSM in Delft.

151