Conclusions Future Directions

As each BVMO is limited in substrate specificity, it is crucial to have a large collection of these oxidative biocatalysts available. Except for expanding the scope of possible reactions, a large toolbox of BVMOs would also increase the chance of being able to perform any wanted specific chemo-, regio- and/or enantioselective reaction. This contrasts with the present situation as only a relatively small number of BVMOs can be exploited for biocatalytic purposes. Therefore, it is still crucial to discover or engineer BVMOs with novel biocatalytic properties.

An obvious way to generate new BVMOs is by taking advantage of the genome sequence information: genome mining. Table 2 illustrates that many putative BVMO genes can be identified which can be explored for their catalytic potential. Two examples of such a genome mining approach have already been reported yielding two novel BVMOs with interesting biocatalytic properties (see earlier17 44). However, it also shows the risks of such an approach. Of the six putative BVMO genes from M. tuberculosis that were cloned, only four resulted in significant expression of the corresponding protein.44 Of the four expressed proteins, only one showed an interesting enzymatic activity. These are complications inherent to genome mining. It is still impossible to predict if a gene can be easily expressed in a certain host nor what specific activities it will exhibit. Nevertheless, as hundreds of unexplored putative BVMO genes can be identified in the genome database, it is attractive to delve into this wealth of sequence information. An effective way to clone novel BVMO genes from sequenced genomes is by using gene synthesis. This technology is becoming relatively inexpensive and allows, e.g., codon optimization for specific hosts. It also circumvents the need to obtain a specific organism or its genomic DNA. By this, obtaining novel BVMOs from sequenced genomes will become more efficient and will yield new BVMOs in the near future.

The recently developed methodology for enzyme discovery that is based on "random" DNA isolation and subsequent screening (metagenome mining) is not yet applicable for BVMO discovery. So far, no screening method has been described that is effective enough to allow screening of massive metagenomic gene libraries. As soon as this technological hurdle has been crossed, the number of available biocatalytically relevant BVMOs will grow.

Another more sophisticated way of obtaining novel BVMOs is to redesign a specific BVMO in order to tune catalytic properties. Such an approach would ideally yield biocatalysts tailor-made to perform any wanted reaction. Random mutagenesis methods have been very popular in the last decade to obtain enzyme variants with improved biocatalytic properties. However, these directed evolution methods typically involve creation of huge libraries of enzyme mutants that have to be screened for a newly introduced characteristic. As a consequence, efficient screening techniques are required to fully screen these libraries. However, methods enabling ultra high-throughput screening are often unavailable and as a result only relatively small mutant libraries are screened. This limits the extent to which enzyme properties can be changed. One cannot expect that by introducing a few random mutations, enzymatic properties will fundamentally change.63 The recently reported directed evolution study on CHMO nicely exemplifies these limitations. As discussed above, screening of a CHMO mutant library for mutants that exhibit altered enantioselective behavior is currently feasible only at low throughput (800 clones per day). A relatively small library (10 000 clones) was screened for CHMO mutants with altered enantioselectivity when converting a ketone or a sulfide.54 The wild-type enzyme already showed some enantioselectivity on both test compounds and therefore the desired change in enzyme reactivity represent energetically seen a small change. Nevertheless, the approach was very successful as several dozens of mutants were found displaying improved enantioselectivity. Several mutants were sequenced and allowed identification of the replaced residues. The mutants represented single, double and triple mutants, indicating effective error-prone PCR. One residue was found in both screens and suggests that this is a hotspot in tuning enantioselective Baeyer-Villiger oxidations and sulfoxidations. As a structure of a sequence-related BVMO, PAMO (40% sequence identity), is available at present, it is now possible to build a homology model structure of CHMO and locate the observed random mutations. We have recently built such a CHMO model structure (unpublished results). Interestingly, inspection of the model reveals that most of the observed mutations are clustered in a specific part of the structure located at the re side of the flavin cofactor (Fig. 6).

When disregarding mutations of surface residues in double or triple mutants, only one mutation was found in the NADPH-binding domain, three mutations were located in the helical domain while all other (10) mutations were part of the interior of the FAD-binding domain. All the observed mutations in the FAD-binding domain are in loops and are close to the predicted substrate-binding pocket next to the flavin.61 Mutations in the same structural region in PAMO introduced by Bocola et al.62 confirmed that this is a hotspot for affecting substrate specificity and enantioselectivity. These results attest to the effectiveness

Figure 6: Model structure of CHMO. The FAD cofactor is shown as sticks. Mutations observed in a directed evolution study resulting in altered enantioselectivities are highlighted (spheres).54 Note that a number of mutations were found in double/triple mutants and cannot be linked with certainty to the altered catalytic properties. Mutations in double/triple mutants that are on the surface were omitted.

Figure 6: Model structure of CHMO. The FAD cofactor is shown as sticks. Mutations observed in a directed evolution study resulting in altered enantioselectivities are highlighted (spheres).54 Note that a number of mutations were found in double/triple mutants and cannot be linked with certainty to the altered catalytic properties. Mutations in double/triple mutants that are on the surface were omitted.

of directed evolution not only to generate valuable new biocatalysts but also to identify residues or structural regions that are important for enzyme functioning. By combining the information obtained by mutagenesis studies and the available structure of PAMO, more directed enzyme redesign studies can be performed in the near future. Such structure-inspired enzyme redesign efforts should effectively generate BVMO variants performing reactions that are yet inaccessible. As CHMO and CPMO homologs are closely related to PAMO at protein sequence level, it is feasible to build structural models of these biocatalysts. Except for tuning substrate specificity or regio- and/or enantioselectivity, it should also be possible to, e.g., engineer mutant enzymes that accept NADH as coenzyme. By solving the structure of PAMO, the residues interacting with the 2'-phosphate of NADPH have also been identified. This could fuel biocatalytic applications that involve usage of isolated enzymes. Generation of a BVMO that can both use NADH or NADPH might also be beneficial for biocatalysis based on whole cells.

Except for extending the toolbox of type I BVMOs, it can also be worthwhile to explore other types of Baeyer-Villiger catalysts. As mentioned above, type II BVMOs have hardly been explored and new types of BVMOs have been discovered in recent years. Furthermore, Baeyer-Villiger oxidation activity has also been introduced into enzymes that normally catalyze other (hydrolytic) reactions.64 This indicates that the field of BVMO discovery and engineering is still expanding and it is expected that the number of biocatalytic applications based on these oxidative biocatalysts will grow accordingly.

Was this article helpful?

0 0

Post a comment