Exploring the limits of vanillyl-alcohol oxidase

Vanillin is the world's principal flavoring compound, extensively used in food and personal products. The curing process of vanilla beans is labor-intensive and the Vanilla plant only grows in a few territories over the world, making synthetically produced vanillin far cheaper than natural vanillin. Nowadays, only 0.5% of the total market is met by extraction of Vanilla beans. The remaining 99.5% of the vanillin produced is of synthetic origin. However, with the increasing interest in natural products alternative methods are being developed to produce natural vanillin from sources other than Vanilla planifolia . Cell cultures, microorganisms, and isolated enzymes form potentially alternative sources for the production of vanillin from natural feedstock.

The Ph. D. project described here was initiated in the framework of the Innovation Oriented Research Program (IOP) Catalysis of the Dutch Ministry of Economy Affairs. In the Enzymatic Oxidation cluster of this research program the catalytic potential of oxidative enzymes for the production of valuable compounds was investigated. Enzymes are an almost unlimited source for the production of these compounds as they can produce natural products and are often highly regio- and/or stereospecific. Furthermore, biocatalytic (enzymatic) processes are in general environmentally friendlier than chemical processes. In this project, we aimed to enlarge the catalytic potential of the flavin-containing enzyme vanillyl-alcohol oxidase (VAO). To that end, the VAO-mediated production of natural vanillin and optically pure aromatic alcohols was addressed. Two different methods were used to direct the reactions to the most favorable product. In the first method we controlled the reaction medium and in the second method we introduced a few subtle changes in the enzyme. For these studies insight in the protein-flavin and protein-protein interactions were of crucial importance.

Enzymatic production of natural vanillin and optically pure alcohols

Chapter 2 describes the VAO-catalyzed conversion of creosol and vanillylamine to vanillin. The enzymatic conversion of creosol proceeds via a two-step process in which the initially formed vanillyl alcohol is further oxidized to vanillin. The production of vanillin is not optimal due to the competitive binding of creosol and vanillyl alcohol in the enzyme active site and the fact that creosol forms a non-reactive covalent adduct with the flavin cofactor.

The oxidation of vanillylamine to vanillin proceeds readily at pH 10. However, as vanillylamine is too expensive for industrial use, we searched for a natural precursor compound. Capsaicin from red pepper is rather cheap and can be hydrolyzed enzymically to vanillylamine by a carboxylesterase from liver or chemically at basic pH values. Therefore, the use of capsaicin as feedstock for the production of vanillin is very promising.

VAO is active with a wide range of 4-alkylphenols bearing aliphatic side chains up to seven carbon atoms. In Chapter 3 , we describe the enzymatic conversion of short-chain 4-alkylphenols to optically pure aromatic alcohols and the conversion of medium-chain 4-alkylphenols to aromatic alkenes. The VAO-mediated hydroxylation of 4-alkylphenols is highly stereospecific (enantiomeric excess = 94%), and the enantiomeric excess of the R -product is even increased by the VAO-mediated oxidation of the ( S )-isomer of the alcohol. The enzymatic dehydrogenation of medium-chain 4-alkylphenols is also stereospecific, suggesting that the p -quinone methide intermediate products are bound in a fixed orientation in the enzyme active site. Some medium-chain 4-alkylphenols are dehydrogenated to the cis -isomer and others to the trans -isomer of the alkene product. Thus, the specificity of the VAO-mediated conversions is dictated by the intrinsic reactivity, water accessibility, and orientation of the enzyme-bound p -quinone methide intermediate.

Tuning the product specificity

In the following chapters (4-7), we studied the possibilities to direct the VAO-mediated conversion of 4-alkylphenols into the most favorable direction using two different strategies. First, we varied the medium in which the reaction was performed and second, we modified the protein by rationale mutagenesis. In Chapter 4 , we investigated the reactivity of VAO with 4-alkylphenols in the hydrophobic solvent toluene and the hydrophilic solvent acetonitrile. In both solvents the efficiency of substrate hydroxylation decreased compared to aqueous conditions. This effect on the hydroxylation efficiency was dependent on the water activity, but independent on the solvent used. This shows that the availability of water determines the efficiency of the hydroxylation reaction. A similar result was obtained by the addition of the monovalent anions chloride, bromide, or thiocyanate. The binding of these ions near the flavin prosthetic group inhibited the attack of water to the enzyme-bound quinone methide, providing a similar effect as lowering the water activity.

Protein engineering of VAO by site-directed mutagenesis proved to be another method to tune the reactivity of VAO with 4-alkylphenols ( Chapter 5 and 6 ). The catalytic center of VAO harbors an acidic residue (Asp170), which is located in the proximity of the flavin N5-atom (3.6 Å) and the substrate Cα-atom (3.0 Å). The location of this residue is intriguing as in most flavin-dependent oxidoreductases of known structure the flavin N5-atom contacts a hydrogen bond donor rather than an acceptor. Asp170 appeared to be crucial for the activity of VAO, the efficiency of hydroxylation of 4-alkylphenols, and the covalent binding of the flavin. Studies from site-directed mutagenesis and protein crystallography showed that Asp170 raises the oxidative power of the flavin cofactor and, therefore, the activity of the enzyme. Replacement of Asp170 by Ser or Ala resulted in a better hydroxylation efficiency of VAO, whereas the Asp170Glu replacement decreased the hydroxylation efficiency. These changes in product specificity are caused by steric effects. The small side chains of Ser170 and Ala170 increase the accessibility of water to the enzyme-bound p -quinone methide intermediate, whereas the more bulky side chain of Glu170 protects the quinone methide from water attack.

In Chapter 7 , we describe the inversion of the stereospecificity of VAO by protein redesign. The active site residue Asp170, involved in water activation, was transferred to the opposite face of the substrate binding pocket (Thr457Glu mutation). As a result, the double mutants D170S/T457E and D170A/T457E hydroxylated 4-ethylphenol to the inverse enantiomer of the aromatic alcohol. This change in stereospecificity is caused by the activation of a water molecule, attacking the p -quinone methide, positioned at the opposite face of the substrate compared to wild type VAO. Crystallographic data confirmed that the distinctive properties of the redesigned mutants are caused by the selective mutations and not by structural changes within the protein. This is the first example of the inversion of the stereospecificity of an enzyme using a rationale redesign strategy.

Rationale of covalent flavin binding

The reason of covalent flavin binding in flavoenzymes is still a matter of debate. It has been suggested that the covalent interaction might a) increase the protein stability, b) enhance the enzyme activity, c) prevent flavin dissociation, and d) improve the resistance against proteolysis.

In Chapter 8 , the role of the covalent protein-flavin interaction was studied by changing the residue to which the flavin is linked. The non-covalent VAO mutant H422A firmly binds the FAD cofactor, but the activity of the enzyme is decreased ten-fold. The lower enzymatic activity is not caused by structural changes but can be fully attributed to the decreased redox potential of the flavin cofactor. Thus, the covalent flavin bond is essential for the high oxidative power of the enzyme.

Oligomeric structure of VAO

At neutral pH, VAO predominantly forms homooctamers. The crystal structure of VAO has revealed that the octamer can be described as a tetramer of dimers in which each dimer is stabilized by extensive intersubunit interactions. Because some dimers are present at neutral pH and low ionic strength, it was of interest to study the stability of the protein assembly as a function of pH by electrospray ionization mass spectrometry ( Chapter 9 ). At low pH values, the octamer-dimer equilibrium shifts to the dimeric form, whereas at neutral pH the enzyme is mainly present in the octameric form. Interestingly, also higher oligimerization assemblies of VAO were observed, indicating that weak interactions between the octamers exist. This information about the oligomeric structure of VAO is very useful for further studies, directed towards the stability of VAO under operational conditions. It is the first time that the mass of such a large molecule (larger than 1 million Da) is determined using this technique.

Conclusions

This research project was performed within the framework of the Enzymatic Oxidation cluster of the Innovation Oriented Research Program (IOP) Catalysis, funded by the Ministry of Economy Affairs. The aim of this cluster was to develop processes for the production of pharmaceuticals, fine-chemicals, and flavors and fragrances using oxidative enzymes, like heme peroxidases, vanadium peroxidases, and flavin-dependent oxidases.

In this thesis work, we focussed on the catalytic potential of the flavoprotein vanillyl-alcohol oxidase (VAO). VAO is active with a wide range of phenolic compounds and can produce a variety of industrially relevant products like vanillin and optically pure aromatic alcohols. We have demonstrated that the reactivity and selectivity of VAO can be modulated by medium engineering and protein engineering.

The principal component of red pepper, capsaicin, proved to be a promising candidate to produce natural vanillin using a bi-enzyme system, consisting of VAO and a hydrolase. By combining these two enzymes a one-pot conversion from capsaicin to vanillin can be realized. This production method yields natural vanillin, which is more valuable than synthetic vanillin. Moreover, the enzymatic production has, in general, environmental advantages compared to the traditional synthetic vanillin production.

VAO produces optically pure aromatic alcohols from 4-alkylphenols. The efficiency of substrate hydroxylation can be tuned by varying the availability of water in the catalytic center or by substituting a single amino acid residue (Asp170) in the enzyme. Furthermore, we were able to invert the stereospecificity of VAO by relocation of the active site base. This demonstrates that protein engineering is a powerful tool to introduce new enzyme characteristics. A major goal for further research would be to enlarge the substrate scope of VAO and to improve the catalytic performance of VAO variants. Interesting target compounds are creosol and capsaicin, as being precursors of vanillin, and epinephrine analogs. Here, random mutagenesis and/or gene shuffling are attractive approaches, since the required changes in VAO are not easy to predict.

VAO is active over a wide pH range, but the protein assembly falls apart under extreme conditions. For possible future applications of the enzyme it is important to study the relationship between the conformational stability and oligomeric structure of VAO. In this aspect, the influence of the covalent flavin linkage is of importance as well. An interesting option to obtain a protein with improved stability properties would be gene shuffling between VAO and a homolog from a thermophilic organism. Another possibility would be a combination of directed evolution methods.