Skip to main content

Effective redesign of oxidative enzymes for green chemistry

Final Report Summary - OXYGREEN (Effective redesign of oxidative enzymes for green chemistry)

Executive Summary:

Life is based on the ability of living systems to perform an enormous array of chemical reactions. This ability is made possible by the existence of enzymes, which are proteins that are capable of accelerating (i.e. catalyzing) chemical reactions achieving rates that are simply beyond the limits of “classical” chemical methods. Such a terrific catalytic power is coupled to an exquisite degree of specificity: enzymes are extremely effective in dealing only with certain molecules and in avoiding the generation of unwanted byproducts. These features make enzymes obviously attractive for industrial applications.
Among the most widely used and relevant chemical reactions are processes that make use of molecular oxygen. We are used to think of oxygen as the molecule that enables respiration. However, one must keep in mind that oxygen is one of the most widely used natural “oxidants” (i.e. molecules that “extract electrons” from another molecule) at the heart of a plethora of most diverse chemical reactions. With the help of oxygen, every organic molecule can be modified. However, it is still a challenge to find a proper catalyst that catalyses such oxidations in a selective manner, without catalysing side reactions while functional at industrial conditions. Enzymes show great promise as selective oxidation catalysts. To exploit oxygen-aided reactions in industrial processes, reliable and robust oxidative enzymes are in high demand.
The goal of OXYGREEN is to gather knowledge on the functions of selected enzymes that use oxygen to perform industrially relevant reactions. This knowledge is intended to be applied for the development of enzyme tools of interest for industrial and biotechnological applications. The project is based on three main cornerstones: (i) biochemistry to understand enzyme function; (ii) genetics and microbiology to modify (‘engineering’ in technical jargon) the biocatalysts of interest as to perform the desired reactivities; and (iii) biotechnology and chemistry to develop the technical tools for the application and usage of the modified biocatalysts and make them attractive for end-user industrial applications.
The project has delivered highly valuable results on all three aspects. Concerning an improved knowledge on how enzymes function, groundbreaking insights have been obtained. This enables us to generate a realistic movie of how an enzyme is able to perform selective oxidations at atomic level. By building several databases (assisting in choosing the right enzyme or enzyme engineering technique) and methods (a number of new methods have been develop for fast and reliable modifying & testing of enzymes), the efficiency of enzyme engineering has significantly been improved. The combined research activities have generated a large set of newly discovered and engineered enzymes. A number of these new biocatalysts have been shown to be successful in an industrial setting.

Project Context and Objectives:
It is a known fact that our way of living must undergo significant changes to face the challenges that the end of the oil era will bring about. These changes must encompass a complete re-thinking of chemistry in terms of novel starting material, technologies, and products, with sustainability and renewability being the keywords. Specifically, we must learn to provide the same commodities as available today (from fuels, textiles, and packaging materials down to fertilizers, food additives, and drugs) or to develop novel, equivalent ones making use of novel materials that come from renewable resources. Several complementary approaches and disciplines are necessary to tackle these challenges, but biotechnology plays a pivotal role because it is built upon millennia of natural evolution towards the transformation of renewable resources into a complex variety of natural compounds with various function and complexity.
The application of biotechnology at industrial scales suffers from two main drawbacks. First of all, natural evolution has not evolved its tools (the cell and the catalytic enzymes contained in it, overall termed “biocatalyst”) for the conditions necessary for technical implementation of the biocatalyst under process conditions. Even though biotechnological tools and technologies to evolve a biocatalyst towards a desired process (e.g. genetic engineering and directed evolution) exist, this has to be done for every desired process and becomes more complex the more complex the chemistry to be catalysed is. Secondly, the chemical compounds that are required to produce all these commodities may not directly be available in nature, in which case it is necessary to rewire the biocatalyst to perform its chemistry on these non-natural substrates and products. The OXYGREEN project has delved right into the heart of both these issues. Its goal was, through a widening of our scientific knowledge on enzyme function (in isolated form as well as in living cells) and of our enzyme engineering toolbox, to bring to a technical level a highly challenging family of biocatalysts with different problems but all sharing the capability to perform interesting chemistry (formation of carbon-oxygen bonds).
Three main families of biocatalysts were addressed; each one with its unique capabilities and each one with a different starting point with respect to a possible final application. Because of this, not all three of them could be brought to the performance level required for industrial application.
In one specific case, the KGDOs, the interaction between biocatalysis and cell physiology was investigated extensively and revealed non-obvious interconnections which paved the way for a holistic engineering approach. The resulting whole-cell biocatalyst was able to meet the industrial requirements for a process aimed at the production of a specific target compound. This bioprocess was transferred to the industrial partner REX and is currently being tested on pilot scale. It will soon result in a tangible impact on people’s lives by enabling the efficient production of (novel) chemical commodities in a more sustainable and environmentally friendly way. Such a successful transfer of knowledge and technology between academia and industry sets an example and model for the future. Undeniably, the facilitation of knowledge and technology transfer between academia and industry, and the bridging of the gap between the two is a pivotal objective for the attainment of the scientific and technological maturity that must accompany the socio-economic transition of this decade.
In the other two cases, Baeyer-Villiger monooxygenases (BVMOs) and mammalian P450 monooxygenases, the starting point was rather far from a final application. Nonetheless, a tremendous amount of knowledge, e.g. on substrate scope, and tools to handle it profitably (such as software and databases), were generated. A significant effort was also made towards demonstrating and expanding the catalytic potential of these biocatalysts, thus promoting interest at the industrial level. In the specific case of P450s, the protein engineering effort led to the identification of several muteins (proteins containing mutations) with improved properties (e.g. better activity, wider substrate scope, different product spectrum). A variety of host microorganisms were tested and significant improvements were made towards efficient whole-cell biocatalysts. A two-phase bioreactor set-up using whole cells of Yarrowia lipolytica was successfully characterized, setting the ground for future studies and improvements. Finally, BVMOs were fused to phosphite dehydrogenase, which allows them to be completely catalytically self-sufficient by simple addition of the unexpensive compound phosphite. Storage and stability studies led to the formulation of freeze-dried enzyme preparations which can be used as an “off-the-shelf” chemical for easy and quick activity screening. A comprehensive ranked database reporting performance of several BVMOs towards a range of substrate and product classes was generated and will greatly simplify future screening and engineering efforts for this enzyme class. Specific reactions leading to products with potential commercial interest have also been identified and demonstrated. Finally, a screening platform for the general purpose of whole-cell activities with NAD(P)H utilizing enzymes was established and validated. The critical momentum that this knowledge and technology are engendering has already sensitized the scientific and industrial community towards the applicative potential that all these biocatalysts hold and will bring more fruits in the near future.

The OXYGREEN-project aimed at obtaining simple, novel, and faster enzyme redesign approaches by advancing methods for diversity generation and knowledge-driven approaches in sequence-alignment and structure-guided mutein (=mutant enzyme) design. Novel screening technologies allowing accurate measurements and mutein evaluation in high-throughput will lead to a breakthrough in biocatalyst platform development over a short time scale.
The research has been organized in 5 distinct but interconnected workpackages. Each workpackage had its own targets and the generated data, insights or methodologies have been input for other workpackages. While workpackages 1 to 4 mainly focussed at knowledge generation and methodology development, workpackes 5 put the generated techniques and enzymes into practice. This workpackage evaluated newly generated biocatalysts and aimed at optimizing methods for employing enzymes/cells for biotechnological processes. Below an overview of the overall objectives for each workpackage is presented.

Workpackage 1: This workpacke targets generation of knowledge that is essential for more efficient enzyme engineering approaches. It involves structural analyses, bioinfor¬matics studies and development of computational tools. This research will deliver improved knowledge and insights concerning the molecular functioning of mono¬oxygenases and the generation of a database that can be consulted for guiding enzyme engineering research, to be used to generate high quality mutant libraries.
Elucidating structures of flavoprotein monooxygenase will shed light on the molecular mechanism of oxygen activation. Determining structures of flavoprotein monooxygenases containing a bound coenzyme and/or substrate will provide new information that can be exploited to pinpoint residues that determine the plasticity of the active site and its ability to stabilise reaction intermediates. The generated database (MutienDatabase) will incor¬po¬rate all relevant structural, sequence, and catalytic properties of generated mutant enzymes.
Workpackage 2: This workpackage has as goal to develop methodologies for constructing and analysing high quality enzyme mutant libraries. Efficient generation of designer enzymes depends on effective methods to prepare enzymes collections within reasonable time and of a good quality. Such methodologies largely depend on efficient and generic molecular biology methods. The method that Oxygreen targets for optimisation is the newly developed SeSaM methodology. As this method is partly based in artificial DNA building blocks, it is significantly different from other existing mutagenesis methods. Thereby it can develop as a superior method.
Except for developing a mutagenesis method, it will also be used and compared with commonly used methods. Furthermore, this workpackage aimed at maturing a bioinformatics tools that allows to compare different mutagenesis methods. The latter tool is now accessible at http://map.jacobs-university.de/map3d.html.
Workpackage 3: This workpackage aims at developing new screening methods for efficiently identify interesting enzymes in mutant libraries (created by WP2). Screening methods for all classes of oxidative enzymes that are part of OXYGREEN R&D should be established. Specifically, P450-based electrochemical screening and conversion analysis methods, Baeyer-Villiger activity screening methods, and product detection methods for KGDOs should be developed. To be able to screen large numbers of enzyme mutants, the methods should be adaptable for high-throughput analyses.
Workpackage 4: In this workpackage innovative high-throughput screening technologies for oxygenase-containing whole cells will be established. There have only been a few methods available that allow quick identification of relevant oxidative activities in whole cells, while such activity is of high interest for industrial applications. Not only for identifying new promising enzymes in mutant libraries but also for optimizing industrial fermentative processes such methods would be highly valuable.
Workpackage 5: WP5 is primarily aiming at evaluating and applying designer enzymes. Knowledge and experience that is generated through these studies will be used as feedback to the other WPs to improve enzyme engineering protocols. Further¬more, more importantly, the use of OXYGREEN-generated designer enzymes for industrial processes will be tested. As part of such application oriented studies, conditions and methods for storing and employing biocatalysts will be optimized. In vitro studies for determining specific catalytic enzyme properties will be performed, but also performance in whole cells will be performed.

Project Results:
The research efforts in the OXYGREEN project have resulted in:
- New knowledge and computational tools to assist in enzyme engineering
We have performed in-depth analyses of a number of oxidative enzymes using a variety of tools that are available to modern biochemists. This part of the research (workpackage 1) has focused on (1) elucidating new structure of enzymes at atomic resolution by using X-ray crystallography, and (2) determining the catalytic properties of the studied enzymes with dedicated techniques.
Enzyme structures - In total, more than 10 new protein structures have been elucidated and deposited at the public protein structure database (Protein Data Bank at www.rcsb.org). The newly obtained structures have been analysed in combination with the determined catalytic properties. The integrated research approach has led us to have an in-depth description of how an oxidative enzyme functions at atomic level. In essence, we now can visualize atom by atom how enzymes are capable to use oxygen to modify the molecules they act upon. Based on the generated data, we have also been able to produce a movie intended for educational and teaching purposes that highlights in a fancy attractive way the progresses and advancements generated by our studies. Furthermore, it is worth noting that, as part of the X-ray analysis studies, we also have successfully used a newly established facility at the ESRF synchrotron facility: a combined analysis by X-ray and microspectrophotometry on one single enzyme crystal. This feat has been highlighted (in ‘ERSF highlights 2011’) and used for a publication in a high impact scientific journal.
Except for a better view on the molecular functioning of individual enzymes, the obtained biochemical information and computational approaches have helped in identifying new biocatalysts from genome sequence databases. This allowed the identification of dozens of new enzymes that can be used for oxidation reactions.
MuteinDB: a database of mutein information - Except for generating new detailed knowledge on enzyme functioning, we also have worked towards a better access to available knowledge. We have developed new tools to allow researchers to effectively and quickly grasp the information generated by our and other studies. This is a key issue because, as common in modern biology, the amount of the data is so huge that they have to be stored and organized in proper way as to make them accessible and useful. More specifically, in our project we have constructed a web-based tool that allows any user to be supported in addressing questions such as: I am interested in a certain oxidative reaction, are there enzymes or modified enzymes useful for this purpose? I am using a certain enzyme, what are the “mutants” (i.e. modifications) that have been already generated and could they be useful for my goals? How should I modify the enzyme in order to make it more effective in performing a desired reaction? These tools are free to use and can be accessed via the internet. The specific tool for gathering information on mutant enzymes that have been studied before can be accessed through the Oxygreen-built MuteinDB (https://muteindb.genome.tugraz.at/‎).

- New methods for creating enzyme libraries
Workpackage 2 of the project aimed at developing generally applicable methods and structure-guided strategies to generate high-quality mutant libraries for accelerated generation of redesigned industrial biocatalysts. The generation of enzyme mutant libraries, an essential step in enzyme engineering, is still a time consuming activity. The R&D community would benefit from improved methods concerning the efficiency of preparing such a libraries and/or improving the quality of a library.
Optimizing SeSaM technology - One focus of this workpackage was improving the SeSaM technology, which enables generation of high-value mutant libraries. The sequence saturation mutagenesis (SeSaM) method has been introduced as conceptually novel and practical simple method for random mutagenesis. SeSaM overcomes most limitations of the often-used error-prone PCR (epPCR) based mutagenesis methods. SeSaM targets in contrast to epPCR each nucleotide “equally” avoiding mutagenic hot spots, achieving subsequent mutations in a codon (up to 37 %), and allowing to adjust mutational biases through employed universal bases. SeSaM comprise four steps: step 1, a DNA fragment pool with random length is generated by the incorporation of nucleotides with phospohorothiolated cleavable bonds; step 2, the enzymatic elongation of the single stranded DNA fragments with the universal base occurs; step 3, the DNA fragments are elongated to full length gene, and; step 4, the universal base is replaced by unmodified nucleotides.
Within the project we worked on optimizing the SeSaM technology. First, the SeSaM method has been advanced to generate protein sequence space that is unobtainable by epPCR. The new SeSaM-Tv(+) protocol and the use of a novel DNA polymerase (3D1) quadrupled the number of transversions, by doubling the fraction of consecutive mutations (from 16.7 to 37.1%). About 33% of all amino acid substitutions observed in a model library are rarely introduced by epPCR methods, and around 10% of the clones carry substitutions that are unobtainable by epPCR. In addition, random mutant library and site saturation mutagenesis libraries have been generated for engineering a DNA polymerase with high read and write functions for the universal base dPTP. For SeSaM it is essential to use such a promiscuous DNA polymerase that accepts such an unnatural base. Furthermore, we have explored a new type of base (purine-like) as component of SeSaM to complete the SeSaM-toolbox. To further increase the amino acid sequence space that can be covered by the SeSaM method, we investigated the inclusion of ribavirin as complementary universal base to the employed dPTP (Figure 1).

The nucleoside ribavirin (dRTP) is a well-known antiviral drug for hepatitis C. The base functions as purine analog and pairs thus with T and C. Novel nucleotide analogs have been prepared and were investigated for the applicability in SeSaM. It was shown that dRTP can be added to the 3´-end of a fluorescence labeled oligonucleotide by the enzyme terminal transferase and represents a “better substrate” for the terminal transferase than dPTP. Furthermore, the engineered DNA polymerase, compared to Taq-polymerase and vent (exo)-polymerase, is able to elongate a mismatch. With the implementation of these improvements, the advanced SeSaM protocol (named SeSaM-P/R) allows for the first time to introduce at all possible positions transversion mutations, which are homogeneously distributed over the targeted gene. With this, the method comes close to the perfect method as it allows altering each nucleotide of a gene. Advancements compared to SeSaM-P lie in the more diverse fragmentations patterns, omitting the use of template-DNA, doubling the mutation frequency, and chemical diversity due to parallel employment of the P- and R-base. The SeSaM-P/R method nearly doubles the number of by epPCR unobtainable amino acid substitutions when compared to previous SeSaM methods. Transversions cause amino acid substitutions that are naturally not or barely occurring leading to chemically diverse amino acids substitution patterns. The latter mutational bias might become beneficial for protein reengineering challenges such as improving solubility of proteins in water, increasing thermal resistance or stability in organic solvents. Summarizing, the SeSaM-technology was successfully optimized towards minimization of unmutated sequences, more transversions and more consecutive mutations in resulting SeSaM-libraries.
Using the current state-of-art in SeSaM, mutant libraries have been prepared for project partners in the OXYGREEN consortium. SeSaM libraries of a Baeyer-Villiger monooxygenases and a cytochrome P450 monooxygenase were created and screened for desired enzyme activities. One of the monooxygenase libraries was used for the benchmarking the advanced SeSaM when compared with a standard epPCR method. To evaluate the monooxygenase libraries, ~500 clones of each library were sent to sequencing to investigate the generated exchanges on the nucleotide level. In addition, ~5000 clones were screened for improved testosterone hydroxylase activity to identify improved monooxygenase variants. As first criteria, the mutational spectra obtained by the two mutagenesis methods were compared. It is known that epPCR libraries generated by Taq polymerase contain a high fraction of AT and TA substitutions. By employing SeSaM that sort of nucleotide exchanges could be reduced. On the other hand, transervsions such as GC that were hardly found for the epPCR were significantly increased in the SeSaM library. By introducing consecutive mutations in a codon the number of possible amino acid exchanges increases. Thus, a larger sequence space can be created by SeSaM. In the epPCR library, no clone (out of 500) was found that was carrying a consecutive mutation. On the other hand, in the SeSaM library 5% of the introduced mutations were consecutive ones. Based on the sequencing results, the SeSaM methodology provides a significantly higher diversity. To finally compare the two methods based on our experience we have set up three criteria: the handling/technical feasibility of the method, the obtainable mutational spectrum and the identification of improved monooxygenase variants. Using these criteria, we conclude that SeSaM is more work intensive in handling than epPCR, but has the advantage of consecutive mutations.
For improving SeSaM, also nucleotide analogs with 3´-protecting groups were tested. Such analogs would afford better control over the addition of universal bases to fragment libraries. Together with the integration of an engineered DNA-polymerase which is able to elongate several base pair mismatches, this would allow the selective creation of subsets of amino acid substitutions. Several desired nucleotide analogs have been prepared by BIL after establishing effective synthesis routes. Also, two different unnatural nucleosides have been synthesized as precursors for subsequent triphosphate syntheses. In one case even the nucleobase had to be built up and attached to the ribose moiety. Nucleosides were phosphorylated to form 5'-O-triphosphates and further modified at the base, at the ribose and the phosphate chain. Diastereomers were separated where applicable. Chemical modifications included protecting groups, conversions from ribo to xylo conformation, introduction of halogens and transient leaving groups.
Out of numerous precursors and intermediates, in total 12 different triphosphates have been synthesized, isolated, highly purified and provided to the consortium partner RWTH. For the consortium partner RUG an immobilisable FAD analogue has been prepared as well. In total ten 3’-modified nucleotide analogs were prepared by the attachment of chemical cleavable blocks with different properties (3’-O-NH2-2’-dATP, 8’-I-dATP, MANT-dATP, 3’-acetyl-2’dRTP, 3’-O-benzoyl-dATP, ANT-dATP, Br-ANT-dATP, Cl-ANT-dATP, Val-dATP, Prop-dATP) and probed in step 2 of the SeSaM method. The MANT and acetyl modifications were not a hindrance for the transferase: several nucleotides could be added to the FITC-labeled oligonucleotide. The other modifications did not work, presumably due to molecular clashes in the employed transferase. Yet, the concept of subsequently attached bases by adding one blocked base, followed by de-blocking and incorporation of a second base and a second de-blocking was not applicable for the SeSaM method. After optimizing the purification of products from the TdT-reaction the yield required for SeSaM was not too low. Nevertheless, the investigations elucidate the reaction conditions required for the transferase-catalyzed reaction and valuable information was gained to advance the SeSaM method.

MAP: a computational tool for decision making - Another activity in work package 2 focused on improving the computational tool Mutagenesis Assistant Program (MAP). The MAP tool was extended in such a way that it assists in generating high quality mutant libraries but also in the preparation of database- and structure-guided enzyme mutants and focussed mutant libraries. Focused BVMO and P450 mutants were prepared and analyzed to fine tune which parts are essential for preparing high-value mutant libraries. Computational algorithms such as MAP are helpful tools to analysis amino acid substitutions patterns of random mutagenesis methods and to develop promising strategies for high quality mutant library generation. The currently available MAP 3D (freely accessible at http://map.jacobs-university.de/map3d.html) provides the probability of amino acid substitutions per amino acid position for 19 mutagenesis methods (Figure 2).
The Mutagenesis Assistant Program (MAP) can now correlate and visualize amino acid substitution probabilities with structural information and will assist experimentalists in the benchmarking of available mutagenesis methods in the future. A protein annotation pipeline included into the Gene Relational Database and linked to the MuteinDB offers now the possibility to align multiple gene sequences of unknown function. By also offering the prediction of 3D-structures, this new tool will assist experimentalists in the construction of structure-guided mutant libraries. Such libraries have been generated for three model genes of this project:
1) Generation and screening of several focused mutant libraries of a Baeyer-Villiger monooxygenase (phenylacetone monooxygenases, PAMO) which revealed six amino acid residues which influence the substrate binding of this monooxygenase. Several positions were identified which in addition to previously targeted positions determine the selectivity of the enzyme and are proposed to interact with the substrate. These residues can be targeted for changing substrate scope of PAMO.
2) The analyses of six structure guided mutant libraries of P450 BM-3 did not only yield improved variants for the mediated electron transfer but also expanded the understanding of cofactor-binding of this enzyme.
3) Based on the retrieved information from the MuteinDB, two sites (E216 and F483) were selected for multi-site saturation mutagenesis in human P450 2D6 (CYP2D6).
Screening resulted in a variant with ~5 fold improved 6β-hydroxytestosterone production and ~25 fold improved production of an unknown hydroxylated testosterone. Docking studies revealed that removing charge from residue 216 and aromatic ring from residue 483 enables more accessible space on sides of the tunnel to active site and lets testosterone to move more freely in the pocket and take conformations horizontal to the heme plane (Figure 3).

Summarizing, all deliverables related to method development for creating enzyme libraries were fullfilled in time and all tasks completed. The advanced SeSaM method now (1) includes new nt-analogs that increase subsequent mutations, (2) allows flexible mutational bias design, and (3) employs an improved polymerase. Two SeSaM protocols (SeSaM-TVII+ and SeSaM-P/R) were published and several SeSaM libraries were prepared and delivered to consortium partners. For the purpose of the SeSaM advancement, not only nucleotide analogs but also different triphosphates with modification on the 3’ position of the sugar backbone were synthesized. Additionally, the benchmarking report on advanced SeSaM and comparison to standard epPCR method was realized. Further, the Mutagenesis Assistance Program (MAP) was successfully advanced to study mutational spectra on protein structures. Generated mutational spectra of the method can now be analysed for local structural components which comprises residue flexibility, solvent accessibility, intra-protein molecular interactions and secondary structural elements. Target based query system has been incorporated to analyse user defined diversity generation method. Subsequently database, sequence alignment and structure guided mutant libraries of P450s, BVMOs and alpha-KGDOs were performed. Generation and screening of several focused mutant libraries of PAMO, P450 BM-3 and P450 2D6 allowed to extend the understanding of the substrate and cofactor-binding and revealed monooxygenase variants with 70 times improved activity for testosterone.

- New methods for screening enzyme libraries
In the logic of the project, the development of tools for creating mutant enzyme collections is paralleled with the development of tools to effectively evaluate the properties of the enzymes and their mutated variants. Most important, a key aim of this part of the project was to develop methods that can be used in a high-through put manner. Only with such methods, large libraries of mutant enzymes can be assayed for desired enzyme activities. The project has yielded several improved or new methods for screening for desired enzyme activities. Generic medium-throughput technologies had to be developed to provide novel screening platforms for oxygenases for instance electrochemical microtiter plate screening for P450 monooxygenases, enzyme-coupled screening for Baeyer-Villiger monooxygenases and a product-based screening for α-ketoglutarate dependent dioxygenases. All these medium-throughput screening technologies had to be developed to enable detection of improved muteins with superior catalytic properties with a throughput of >1000 mutants/day. One key target during development of screening platforms is to liberate cytochrome P450 enzymes from the need of costly coenzymes (NADPH) and to evolve self-sufficient BVMOs. New screening assays and technologies will be used for screening enzyme mutant libraries, including libraries that had been prepared within the project, e.g. SeSaM-based mutein libraries.

New generic activity screening technology for BVMOs - A colorimetric assay based on phosphate production has been successfully developed to screen for BVMO activity. New generic screening platforms will allow effective selection of active or improved BVMOs from mutant libraries. For such approach, the activity of a BVMO has to be translated into a measurable signal. A method that identifies active enzyme based on color formation would be attractive, but little information is available about colorimetric/chromogenic assays for BVMOs. In general, BVMO mutant libraries have been screened either by following the consumption of NADPH in time (at 340 nm) or by gas chromatography. In order to have a faster and generic screening assay, an indirect assay for detection of phosphate was developed applying a phosphite dehydrogenase as reporter enzyme (see figure 4).
This latter inorganic compound is a product of the oxidation of phosphite by phosphate dehydrogenase, an enzyme that can be applied for the regeneration of NADPH. With this assay, 0.3 to 10 mM product formation (i.e. phosphate) can be detected in MTP format allowing a throughput of >1000 mutant per day. Apart from using the phosphate assay with whole cells, the assay was also shown to be useful for screening activity of cell extracts. It was used to perform substrate profiling studies: determining the range of compounds accepted by muteins or newly discovered BVMOs. A panel of 50 representative substrates including industrially relevant compounds was investigated. The end-point assay is very helpful for such screening as it allow visual inspection of substrate profiles of BVMOs or other redox enzymes.
Finally, also the detection of hydrogen peroxide by using the Peroxy Green 1 dye has been implemented in the assay. This allows for the identification of so-called uncoupling mutants which consume NADPH without catalyzing the Baeyer-Villiger reaction and produce hydrogen peroxide instead of valuable target products. By adding this assay method, it is possible to discern false positives.
The phosphate-based screening assay has been described in two scientific publications and has already helped to identify interesting enzymes, see below.

Chromatographic screening method for dioxygenase mutein screening - An analytical chromatographic method based on the LC-MS technology, that enables semi-quantitative medium throughput analysis of product, substrate, and co-substrate (glucose and a-ketoglutarate) of the KGDO-reaction was developed. Furthermore a product-based, quantitative, colorimetric assay was adapted to 96-well plates and made suitable for high-throughput screening of the KGDO-reaction, allowing a screening for new KDGO-muteins in MTP format. The assay is very specific and highly sensitive, as it can detect down to 1 µM product in the linear range of 1 to 40 µM. With this new screening assay in hand the screening of KGDO-libraries generated in WP2 was performed and new muteins with desired properties were found. After cell lysis, the in-vitro screening procedure based on spectrophotometric determination of hydroxyproline was performed to identify optimized P4H variants from SeSAM library. Muteins with different kinetic characters were identified. By applying the same assay, the obtained muteins were characterized with regard to catalytic efficiency, whereas the industrial partner REX tested the new catalysts during up-scaling.

Electrochemical screening technology based on microtiterplates – A main challenge in industrial application of P450 enzymes is their cofactor dependency. A promising possibility to solve this issue is the substitution of the cofactor by electrochemical currents, which includes the use of soluble redox mediators. Such mediators shuttle electrons between an electrode and the enzyme. This can be performed in solution or by immobilization of the enzyme at the electrode surface enabling a direct electron transfer. For both strategies a technology is needed to parallelize electroenzymatic experiments and to accelerate research efforts. In the Oxygreen project an electrochemical microtiter plate (eMTP) was developed which enables parallel electrochemical experiments with simultaneous optical data acquisition in up to eight wells of a microtiter plate (see figure 5).
This is the first example for parallel spectro-electro-chemistry in commercially available microtiter plates, with each well containing a set of independent electrodes. The eMTP allows a fast screening of varying reaction conditions.
The well known redox probe potassium ferrocyanide and the mediator ABTS were characterized in the eMTP and the functionality and reliability of the system were proven. Furthermore a mutein library of P450 BM3, was screened for its performance with the pNCA assay using cobalt sepulchrate as mediator between enzyme and electrode. The eMTP allowed a fast comparison of P450 BM3 muteins in terms of reaction rates. In addition several reaction conditions were varied and the influencing parameters identified. The eMTP was used to find optimized reaction conditions for the electroenzymatic process. For example some muteins were found that exhibited a different potential optimum or displayed an altered mediator concentration dependency. The mediator driven approach was subsequently applied in a small-scale electrochemical reactor to convert p-xylene to the industrially important product 2,5-dimethylphenol with applying a P450 BM3 mutant developed within the Oxygreen project. Conversion of the respective substrate with the improved mutein revealed that the hydroxylation with the newly designed electrochemical set up is possible, therefore replacing the expensive cofactor NADPH in the reaction system.

P450 muteins active with conducting polymers - The screening of the structure-guided P450 BM3 mutant libraries did not lead to activity improved variants with a conducting polymer for the electrochemical conversions. This alternative cofactor system was identified to be not suitable for an electrochemical application therefore the research was focused on the second proposed alternative cofactor system: Mediated Electron Transfer (MET). Previous studies on MET build the proof of concept. The evolution of P450 BM3 M3 reported by Nazor et al. for improving the mediated electron transfer (MET) could further be extended towards increasing activity in presence of the alternative cofactor system Zn/Co(III)-sep. As starting P450 BM3 mutant was evolved in using epPCR and SeSaM for generation of random mutant libraries. The libraries were enriched for active clones with the new high throughput flow cytometer whole cell screening (FACS) system established in within the project, see below. The clones were screened for improved activity using Zn/Co(III)-sep as sole electron source for P450 activity. Improved variants were sequenced and characterized for key catalytic parameters. The best P450 BM3 mutant displayed a nearly 4 fold higher activity (kcat = 6.8 mmol product min-1 mmol-1 P450) compared to the starting variant P450 BM3 DM using the alternative Zn/Co(III)-sep cofactor system. In addition, the catalytic efficiency could be increased by 2.5 fold compared to the starting template. Four new hot spots could be identified, saturated individually and recombined to select the most beneficial amino acid combination for improved MET in P450 BM3. A further improved mutein was found with a 1.4 fold improved activity. Summarizing, the selected improved variants give valuable information, which amino acid substitutions are relevant for MET-based biocatalysis.

High throughput LC-MS analytics for steroid hydroxylation - A high throughput LC-MS screening method of human CYP2D6 based on whole cell conversion was successfully set up using bufuralol, metoprolol, propranolol or testosterone as substrates. The work was the result of an intense collaboration between TUG and VUT.
Conversions of steroids can be performed in 96 well MTP format (µL scale) prior to injection of 5-10 µL for HPLC-MS analysis. The chromatographic method was established for an Agilent 1200 system using the G1956B-MS-module with an ESI spray chamber. For efficient separation a reverse phase column is tempered and the composition of the mobile phase was optimized. To reduce the analysis time for each reaction to < 2 min, a flow rate of 1.5 ml/min is used. For this specific class of substrates, also conversions with microsomal preparations of yeast strains overexpressing CYP2D6 were done. The developed system allows analysis of ~700 samples per day, which is sufficient to screen enzyme mutant libraries with mutations of up to 3 sites.
For some of the envisaged assays described above, dedicated organic synthesis was necessary to prepare the desired assay compounds as well as steroid substrates and standards for analytics. In a first attempt to isolate and characterize the hydroxylated unknown metabolite formed by a CYP2D6 mutant the fermentation broth was analyzed directly without any further concentration step. After centrifugation of the biomass, the supernatant was supplemented directly to HPLC-MS analysis, to identify potential hydroxylated steroid products. An analytical workflow for efficient analysis of unknown steroid products was developed (see figure 6).

Starting from crude sample isolation, followed by an HPLC separation based on MS analysis and product isolation via a solid phase extraction unit, the sample was analyzed using a NMR spectrometer for structural analysis. Hence, concentrated fermentation samples were re-dissolved in a mixture of ACN/MeOH and HPLC analysis was conducted. By using an XB-C18 column and an ion trap mass spectrometer it was possible to identify the three main fermentation products. After identification of the desired hydroxylated metabolites via mass spectrometry, structural analysis via NMR spectroscopy was performed to distinguish between different hydroxylation positions. The major hydroxylated product that we were able to isolate and identify, based on HPLC/MS-SPE/NMR analysis and previously published spectral data were 2,17-dihydroxy-(2β,17β)-androst-4-en-3-one. Additional 2-D experiments (COSY and HSQC) and the change of solvent completed the analysis.

Cell-based screening methods -To generate designer enzymes by protein engineering comprises three subsequent steps: efficient generation of diversity on the protein level, efficient expression of the target enzyme in the expression host and an efficient screening system to identify novel enzymes with desired features. The availability of an appropriate screening system is essential in protein engineering experiments (“You get what you screen for!”) and is thus determining the success of such approaches. Many important redox enzymes need NAD based cofactors, which are efficiently regenerated by the cellular metabolism. In addition whole cells can also act as a cover protecting the enzymes from shear forces or organic solvents and, thus, increase their stability. On the other hand such the protecting cell sometimes limits substrate access to the intracellular enzymes. Another advantage of cell based screenings is that enzyme isolation and purification steps can be circumvented which speeds up the screening process. Therefore to speed up breeding cycles, to reduce costs and to simulate later processes as far as possible, work focused on the development of whole cell screening systems.
The aim was to develop whole-cell based screening systems, which can be employed to screen the three types of monooxygenases investigated in the OXYGREEN project, namely cytochrome P450 enzymes (P450s), Baeyer-Villiger monooxygenases (BVMOs) and alpha-ketoglutarate dioxygenases. While P450s and BVMOs need NAD(P)H as cofactors, the employed dioxygenases depend on ketoglutarate for substrate oxidation.

Whole-cell screening of P450 activities - Different approaches have been followed to set up efficient P450 screening systems. One of these approaches was based on alginate beads as compartments for whole-cell screening. Growth of E. coli and protein expression on the example of the P450 BM3 in these alginate beads were successfully shown. Subsequent use of these beads in an oxygen depletion assay (molecular oxygen is consumed by all P450 enzymes during the reaction) worked well in a macroscopic setup. The transfer of this system to microwell plates faced its limitations in sensitivity due to the low reaction rates of P450s. The yeast Pichia pastoris was too heavy and oxygen demanding for a homogenous distribution in alginate beads. However as demonstrated and published during the Oxygreen project, this yeast was the most efficient expression host for the human cytochrome P450 2D6. Therefore, a novel screening system was developed which was based on whole-cell conversion of the target steroid substrates such as testosterone coupled with analysis HPLC-MS. Thus, a screening system with medium-throughput but sufficient capacity for the later studies was established and used to screen the CYP2D6 libraries.
To further increase the screening throughput, emphasis was put at developing a screening method based on fluorescence-activated cell sorting (FACS). Employing FACS, one can routinely sort > 107 clones or cells per hour. Different approaches such as coupling FACS with oxygen sensitive dyes have been evaluated for technical feasibility. Finally, a whole cell ultra-high throughput screening platform for P450 monooxygenases based on flow cytometry was established that is based on the dealkylation of a coumarin derivative. This platform is not only restricted to P450 monooxygenases activity, but can also find applications in metagenomic screenings, improving protein expression and screening of recombination libraries of P450 monooxygenase as well as corresponding reductases.

Whole-cell screening of BVMO activities - The original plan to set up a FACS-based screening for BVMOs had to be abandoned due to lacking technical feasibility. Instead, an innovative chromogenic assay based on the detection of phosphate formed was established. This latter inorganic compound is a product of the oxidation of phosphite by phosphite dehydrogenase, an enzyme that can be applied for the regeneration of NADPH. A prerequisite for this screening approach was the translocation of the BVMO to the periplasm via the Tat export system. The screening was found to be a powerful tool to screen BVMO libraries as it enables the identification of several new BVMOs.
In addition, using the nanospot technology developed by the Oxygreen consortium supported by a subcontract for library printing on membranes and agar plates, a reliable NADPH depletion assay was set up, which is suitable for screening e.g. BVMO libraries and libraries of other NAD(P)H consuming enzymes.

Whole-cell screening of KGDO activities - Screening of a-KGDOs was based on above-mentioned colorimetric assay after adopting the protocol for whole-cells. Synchronized growth of the cells in the microtiterplate was crucial for the development of a reliable assay.

Concluding, the feasibility and the benefit of the newly developed screening methods is underlined by the fact that these tools have been successfully employed in screening libraries of P450s, BVMOs and a-KGDOs. This has resulted in the identification of improved enzyme variants. Variants of the human cytochrome P450 CYP2D6 displaying higher activities and altered regio-preference in steroid hydroxylation were generated. Unprecedented results have been obtained demonstrating that human P450s can be changed from almost no activity on specific drugs to similar activities as some of the best known hydroxylases. In addition, the bacterial monooxygenase P450 BM3 has been improved for the hydroxylation of industrial relevant substrates; the corresponding mutants have been patented. Furthermore, BVMO mutant were identified which showed e.g. remarkable changes in the accepted substrate spectra. Also for the group of KGDOs, muteins with improved hydroxyproline production rates were identified by using the developed screening method. Thus, the R&D towards developing new screening technology was successfully accomplished and the obtained results will be widely applicable for future enzyme engineering experiments and discussion with pharma companies confirmed interest in the mutants developed as model enzymes of the project. Several of the identified mutant enzymes have been patented or are still being considered for patent applications. Industrially relevant follow-up projects that will exploit the Oxygreen-generated knowledge and methodology have initiated.

- Exploiting the generated knowledge and technologies for the generation of industrially relevant enzymes

As last step towards industrial valorization of the OXYGREEN project output, the generated enzymes have been explored for their performance as industrial catalysts. The research activities for this were organized in workpackage 5. Specific outcome is sketched below.

New P450 monooxygenase-based biocatalysis – Several P450 monooxygenases have been produced and tested for industrially relevant chemical conversions. For effective production of eukaryotic monooxygenases, yeast expression systems have been successfully applied. This has also enable tod explore the use of whole cells for performing steroid oxidations. Hydroxylated steroids could be produced in this way. For hydroxylation of aromatic compounds, the use of whole cells and cell extracts have been tested in an industrial setting. This revealed that the use of cell extracts may be beneficial in the case of hydroxylations of aromatic compounds. P450 monooxygenases generated through the OXYGREEN project were shown to perform well, resulting in a patent application.

New KGDO-based biocatalysis – For the industrial production of a non-canonical amino acid, REX and TUD have succeeded in creating a whole-cell biocatalyst that can be used in a fermentation process. Beside protein engineering, this also involved metabolic and reaction/process engineering. Metabolic engineering aimed at the optimization of P4H catalysis in whole microbial cells focussing on optimized recombinant gene expression and the coupling to central carbon metabolism in terms of cosubstrate (alpha-ketoglutarate) supply and substrate (proline) uptake. Metabolic and reaction engineering targets were identified by physiological characterization of the microbial catalyst and metabolic flux analysis. These analyses gave valuable insight into whole-cell biocatalyst operation and factors limiting enzyme, whole-cell biocatalyst, and process performance. With a metabolic delta-putA mutant, good process performances could be reached. Reaction engineering was based on growing as well as resting cells. Both approaches allowed the development of productive process setups. The first tests in an industrial setting have shown that the academic research can be transferred into a real biotechnological process. This paves the way to a new biotechnological process, based on a whole cell biocatalyst resulting from the OXYGREEN project.

New methods to stabilize, store and apply BVMOs – To establish easy and affordable methods to store and subsequently apply monooxygenases, several methods have been developed.
For providing an easy and cheap solution for the use of coenzyme-dependent enzymes, fused enzymes have been produced. By producing BVMOs fused to an auxiliary enzyme, phosphite dehydrogenase, regeneration of the required coenzyme NADPH can be achieved by using inexpensive phosphite as electron source. This method was shown to work for all tested BVMOs and provides an easy way to test and apply these enzymes.
For generating more stable BVMO biocatalyst, a newly developed computational method was applied for the most widely used BVMO, cyclohexanone monooxygenases. Though this biocatalyst can be used, applications suffer from the poor stability of the enzyme. By applying a careful structural analysis, a relative small set of mutant CHMOs was prepared that should contain variants with improved thermostability. Analysis of 20 mutant resulted in the identification of 3 mutations that lead to improve stability. This has led to the production of a significantly improved biocatalyst and shows that the computational approach is powerful in improving the thermostability of enzymes: an excellent demonstration of the power of using newly developed enzyme engineering methods.
An easy-to-use biocatalyst formulation was developed by using freeze-drying. By careful tuning the composition of the enzyme preparation, extremely stable and functional enzyme formulations have been prepared.

New BVMO-based biocatalysis – The BVMO enzyme class has been extensively studied in academic biochemistry and biocatalysis research and is usually described as highly selective in all chemical aspects. The knowledge gained in previous studies and newly discovered biocatalysts was taken as the basis for this work. The evolution of an industrial process was further divided in three tasks: catalyst discovery and systematic studies, refinement and evaluation by supporting theoretical models and, eventually, catalyst and process engineering.
Task one combined substrate profiling of novel monooxygenases for their performance in Baeyer-Villiger oxidations in parallel reactions on analytical scale and systematic investigations of using BVMO for defined substrate classes. VUT played a major role in this. The enzymes were mostly supplied by RUG and substrate classes were in many cases suggested by DSM to consolidate the application character of these investigations. The former studies provided valuable insight on substrate promiscuity and catalytic performance of new 13 BVMOs and three related monooxygenases, allowing interpretation of biochemical and structural relationships and evaluation of these enzymes as tools in synthetic organic chemistry.
Substrate-specific studies were mainly aimed at industrially relevant intermediates and aroma compounds and thus demonstrated the possibility to apply BVMOs in the synthesis of
- (1) the platform chemical 3-hydroxypropionic acid by oxidation of levulinic acid derivatives,
- (2) Aerangis lactone, a widely used fragrance compound,
- (3) aroma lactones from Mango and Jasmine fruits and plants, and
- (4) monomers for the production of polyamid-9 materials.

The second task dealt with the development of theoretical tools for data organization, evaluation, visualization and interpretation. On this account a relational database for BVMO transformation results was implemented, serving as basis for the development of a quantitative catalyst scoring algorithm with graphic output. The potential of this analysis was illustrated with cyclododecanone monooxygenase from Rhodococcus ruber, a very versatile biocatalyst (Figure 7).
Additionally, a systematic study on the regioselectivity of BVMOs was conducted, ultimately leading to a very accurate descriptive in silico model for Baeyer-Villiger oxidation of diketones; this project could only be realized in collaboration with partner BIB, who supplied the expertise in molecular modeling, protein-ligand docking and in silico enzyme design (Figure 8).
The research culminated in the development of three showcase processes demonstrating the application potential of BVMOs in industrial biotransformations. First, a mutant enzyme is used in the optimized biocatalytic synthesis of the platform chemical ethyl 3-acetoxypropionate at a calculated productivity of 17 g L-1d-1 starting from cheap levulinic acid. The oxidation of levulinic acid or its esters opens up a new pathway for the generation of value-added bulk chemicals from renewable sources. Levulinic acid is generated as a byproduct from carbohydrate dehydratization in significant percentage and also yearly tonnage and is therefore available at a competitively low price. By oxidation levulinic acid, 3-hydroxypropionic acid can be prepared. This latter chemical is currently used as crosslinking agent for polymer coatings and metal lubricants and as antistatic agent for textiles. More importantly, it was mentioned in third place on the US Department of Energy’s current Top Value Added Chemicals from Biomass as sugar-derived building block. This C3 compound has the potential to be a key building block for deriving both commodity and specialty chemicals as 3-hydroxypropionic acid is not part of any current petrochemical-based technology (Figure 9).
Second, natural Aerangis lactone is produced in perfect optical purity via a highly efficient chemo-enzymatic sequence process (121 g d-1), requiring just a single operation step. Aerangis lactone was characterized as the main odor component of African white-flowering orchids and is widely used in the perfume and cosmetics industries. The developed process delivers this important aroma compound with high productivity and selectivity, offering a dramatic improvement over known unselective processes or multi-step non-scalable selective synthesis routes (Figure 10).
Third and last, BVMOs are used as unselective chiral catalysts in the synthesis of a key intermediate in polyamid-9 monomer synthesis. Polyamide plastics (PA) constitute a worldwide multi-megaton market with a yearly turnover of >25bn €. In addition to the major products PA 6 and PA 66, other PAs with different monomer lengths are widely used in high-performance applications. Their properties can be highly superior to their C6-counterparts, but raw material prices and ensuing multistep syntheses often compromise their use on broader scale.
We envisioned introduction of principles of green chemistry into this pathway. These optimizations should focus on the development of a biocatalytic process towards lactone 2a. Substrate profiling of the recently discovered cyclopentadecanone monooxygenase (CPDMO) enabled the possibility to access lactone 2a with synthetically useful productivity of 8.7 g L-1d-1 in a pilot scale biotransformation. Such an enzyme-based process shows great promise to replace hazardous and polluting chemical approaches (Figure 12).



Based on these results a transfer of the technology under industrial relevant conditions was successfully demonstrated.
The above presented cases of BVMO-based catalysis demonstrate the link in this work package between enzyme engineering and biotechnological applications, propagating the use of monooxygenases in fine and bulk chemical synthesis.

The generated S&T results/foregrounds of the OXYGREEN project have been used to set up a website that offers ‘OxyGreen Services’: www.enzymedesign.org. This will facilitate dissemination and usage of all developed knowledge and techniques derived from the OXYGREEN project (Figure 13).
Potential Impact:
The OXYGREEN project aimed at developing new enzyme engineering methodology and industrially applicable enzymes. The resulting enzyme-derived tools will enable the organic chemist and process engineer to integrate a selected oxidative enzyme(s) in their process to make it more cost effective and environmentally friendly. In essence, it is the transfer of the capabilities that living organisms have developed in the course of evolution into processes used by chemical and biotechnology industries. The ultimate was twofold: (i) to advance our current understanding of the functions of the enzymes that use oxygen to perform chemical reactions and (ii) to translate this knowledge into tools and protocols that can eventually lead to improved technologies for obtaining cleaner, cheaper and more robust (safer) industrial processes. The involvement in our project of a few well-known chemical industries assured the translation of fundamental research into new industrial applications.
The tools developed within the Oxygreen project will allow a faster and more purposeful development of monooxygenases in the future. New screening methods for oxygenase engineering became available at the research sites of several Oxygreen partners (e.g. Aachen, Groningen, and Graz) and are now offered as services for future collaboration projects and as service on demand (www.enzymedesign.org). The broadly available technologies facilitate access to other academia and industry compared to exclusive secret know-how developed in some companies and transfer of the Oxygreen results to the European industry. On the other hand several interesting enzymes have been used to develop new generic mutagenesis and screening procedures and the generated improved enzymes can be used in drug development processes as well as in the production of other industrially relevant compounds such as vitamins, polymers and human drug metabolites for toxicology studies and for the development of a new generation of drugs.

Specific details on the impact of the performed research are provided below. Details are provided on (1) creation of new enzymes, (2) development of new technology for enzyme engineering, and (3) applications of enzymes.

(1) The OXYGREEN-project has specifically targeted the design of industrially applicable oxygenases for use as oxidation catalysts. There is a huge demand for such biocatalysts. The focus has been on three different enzyme types: P450 monooxygenases (P450s), Baeyer-Villiger monooxygenases (BVMOs) and ketoglutarate-dependent dioxygenases (KGDOs). A collection of new oxidative enzymes has been created, by using the knowledge and technology that was developed through the project:
* P450s - Engineering the cytochrome P450 2D6 revealed valuable information on essential amino acid residues of this important human liver enzyme, thus gaining new insights. This novel knowledge might be exploited of the development of pharmaceutical active ingredients in the future. The most important results are publicly available via the MuteinDB. Some of these results will change the current opinion on the in vivo function of human microsomal P450 enzymes and this new knowledge will have a significant impact in respect personalized medicine concepts and in general on the development of future human drugs.
The conversion (hydroxylation) of industrial important substrates by engineered P450 BM3 mutants provides products which are precursor for vitamins synthesis, fine chemicals and pharmaceuticals. Advantages are a cheaper, time saving and green production in sometimes more cost effective manner. The OXYGREEN-project has resulted in a large number of improved enzymes, for selective hydroxylations. The market value of these new biocatalysts is shown by the patent activities of DSM (one patent within OXYGREEN). The generated variants will be implemented in production processes at DSM.
** BVMOs – BVMOs have been studied and redesigned with the aim to create a set of useful biocatalysts. This has been shown to be very effective: >40 different BVMOs exhibiting new reactivities have been designed. Evaluation studies have also confirmed that a number this newly created biocatalysts can be applied for industrially interesting reactions. Examples are conversions that lead to polymer building blocks and fragrances. Currently, several of these biocatalysts are being considered for patent applications. A selection of the generated BVMOs are currently also offered via an industrial partner.
*** KGDOs - KGDOs are unique in that its chemistry is directly linked to the cellular metabolism, and as such benefits from the use of whole cells as biocatalysts (rather than isolated enzymes). Through publications and presentations, we successfully endeavored to convey to the scientific community the message that, if we want to develop industrial whole-cell based biocatalysts for challenging and interesting chemistry, we need a holistic (systems biotechnology) approach based on a deep understanding of the relationship between cellular physiology and biocatalytic reaction and considering the enzyme, host, reaction and process levels for truly integrated bio-process engineering. Other groups have begun to recognize the importance of this approach and are shifting towards this path (see latest publications by STL Harrison and PC Cirino). Moreover, the knowledge generated by our research effort has been taken up by the industrial partners of the project, enlarging their toolbox in chemical synthesis, and will result in the development of an efficient, industrial process for the production of one target compound. The impact of such innovation will certainly be significant although difficult to estimate in detail, since the impact on people’s lives depends on the specific commodities that will be produced form such chemical intermediates and will benefit from, e.g. the novelty of a compound and its benefits, a lower production price, and more sustainable/environmentally friendly production processes.

(2) Another cornerstone of the OXYGREEN-project was technology development. This has led to new methods for creating high quality enzyme mutant libraries. A major contribution has been the advancement in the commercially exploited SeSaM method. The method development permits to fasten the screening time and quality and therefore is judged to have a cost effective impact.The directed evolution protocol and the identification of mutations was improved, in terms of quality and in a time saving manner. Through advanced protein engineering methods, proteins can be evolved and improved more effectively and subsequently new catalysts can faster be included in industrial processes.
Another technology development activity has been the work on establishing new method to measure and screen for desired enzyme activities. Such efficient methods are essential for the process of enzyme redesign. For all studied enzyme classes tailored screening methods have been developed. This includes methods that can be applied with whole cells and therefore allow the screening of huge mutant libraries. The success of such methods has been demonstrated by the discovery of new and relevant enzymes. The developed methods in electrochemically driven enzymatic reactions are the first steps to alternative cofactor systems for monooxygenases. The electrochemical driven conversion of industrial important substrate could be demonstrated as proof of concept. Enzymatic reaction driven by electrical current generated through renewable energy is the future perspective to generate in a cost effective and environmentally friendly way high valuable products.
Also several computational tools have been developed. This includes a new database that contains data on known biocatalysts and that can aid researchers for making the right decision on which biocatalyst to use or to engineer. Another computational tool is the MAP server that assists in selecting which enzyme engineering approach should be used for an enzyme of choice. Both computational tools are available free of charge for researchers all over the world. Another output concerning computational research are the protein structures that have been elucidated during the project. All structures have been deposited in the international protein structure database.
Biologics (BIL) has successfully synthesized high purity nucleoside triphosphates with unnatural nucleobases in order to support and improve the SeSam technology. In addition, a large number of rare nucleoside, nucleotide and nucleobase analogues have been prepared as corresponding precursors and intermediates. Initiated by the project the company has also started to build up experience on FAD analogues, a class of compounds that will be another brick within its expanding product portfolio. Most of these new structures have been or will be made commercially available and will considerably strengthen the position of BIOLOG LSI as a source for rare and precious research probes in Europe.

(3) Except for creating new enzymes and technology development, the OXYGREEN has yielded a number of new enzymes that can be applied as biocatalysts in biotechnological processes.
The application-oriented nature of WP5 gave us the possibility to engage in research directly related to manufacture of marketed chemicals. The use of biocatalysts often gives unique advantages over classical chemical production processes in terms of energetic and ecological sustainability; these points could be demonstrated with the development of two exemplary pilot processes based on OXYGREEN-designed BVMOs for the synthesis of a widely-used aroma compound, Aerangis lactone, replacing a hazardous chemical pathway with benign reaction conditions while simultaneously improving the selectivity towards the natural form of the compound. Secondly, with the oxidation of levulinic acid, a by-product of biomass degradation, it could be shown that by means of BVMO-based biocatalysis a new pathway towards the highly interesting platform chemical 3-hydroxypropionic acid was enabled. This building block is difficult to obtain by other means and serves as fossil fuel-independent starting point for many known industrial bulk chemicals. Also a furan-based renewable building block could be converted and the respective process is being considered for a patent application. P450s have been successfully tested for the hydroxylation of aromatic compounds. This has resulted in a patent by DSM. Rexim has targeted the production of hydroxyproline.
Hydroxyproline is a well know amino acid worth an estimated market of several tens tons. Evonik Rexim SAS, which used to be a leader on this market a few years ago, unfortunately had to exit the market after the mad cow disease: this amino acid as all the other used to be extracted from animal protein hydrolysate. In early 2000, animal origin was totally rejected from the market and the production of the main aminoacid was then tranfered to fermentation but the targeted amino acid was no more a possible byproduct. Until now, no efficient hydroxyproline technology was available. This study aimed at developing a route that would be both adapted to industrial possibilities and economically worth the investment. From the results obtained during the evaluation phase, a process emerged that matches industrial requirements and purification is possible from a fermentation origin: the process seems to be feasible on industrial scale even if more optimization would still be necessary. The process envisaged is compatible with the equipment available in the European plant of Evonik Rexim and will be a new product produced with the EU.

The dissemination has been achieved along the lines originally sketched in the proposal of this project:
• Patent protection. To allow European industries to commercialise the results of the work, part of the results have been protected by a patent. Further patents are currently being considered.
• Cooperation with companies. The project has catalysed establishment of collaborations between OXYGREEN-partners and external partners. Several academic partners and industrial partners have assured continued collaboration by follow-up projects. Also, links have been made with SMEs outside the OXYGREEN project, e.g. enzymes that have been generated through the project are currently being sold at an SME.
• Publication in the scientific literature. The project has resulted in an impressive number of publications (>70) and a similar number of posters & lectures at scientific meetings.
• Submission of reports to the EC and participation in technology transfer activities. Reporting has been performed according to the contract. For technology transfer, the PIs of the involved partners have promoted OXYGREEN at numerous meetings that were also attended by end-users.
• Presentation at national and international scientific meetings. The involved responsible scientists are frequently invited to present their work at national and international meetings, which serve as an important platform for transfer of new knowledge. • Teaching. All academic participants have been involved in education of students, both at their own University and as invited speakers at post doctoral courses for persons from industry, government, and regulatory agencies. There has also been a publication, describing an industrial biotechnology experiment, in a journal for promoting education in chemistry. This has been adapted in practical courses at academic institutions throughout the world.
• Follow up with demonstration projects or industrial contract research. Several academic partners are involved in follow-up projects that build on the expertise and enzymes generated through OXYGREEN.
• Exploiting the internet. Several internet-based tools have been developed (MuteinDB, http://muteindb.genome.tugraz.at and MAP-3D, http://map.jacobs-university.de/map3d.html). Furthermore, the OXYGREEN website (http://www.oxygreen.org) has been set up and used. For offering the knowledge and technology available through the OXYGREEN consortium, a dedicated Oxygreen Services website has been put in place (http://www.enzymedesign.org).

List of Websites:
Websites: www.oxygreen.org and www.enzymedesign.org

Contact details:
Prof. Dr. M.W. Fraaije,
Groningen Biomolecular Sciences and Biotechnology Institute
University of Groningen
Nijenborgh 4
9747 AG Groningen
The Netherlands
Email: m.w.fraaije@rug.nl
Phone: +31 50 3634345
Fax: +31 50 3634165