Studies on the influence of loop structures on the selective hydroxylation of sesquiterpenes in Rieske Dioxygenases

ABSTRACT

Oxyfunctionalized compounds play a pivotal role in synthetic chemistry, as flavour- and fragrance compounds, pharmaceuticals or base chemicals. In many aspects of life, oxyfunctionalized molecules play an important part. Especially for fragrances and pharmaceuticals, controlling selectivity in oxyfunctionalization reactions is of high importance, as different isomers can lead to completely different biological functions. Selective monohydroxylation and the cis-dihydroxylation of arenes is still a major issue for organic chemistry. Especially in compounds with multiple unsaturated double bonds, like sesquiterpenes, the regioselectivity control is challenging. Additionally, the transition-metal catalysts used are often toxic, need high temperatures and pressures and the reactions are usually performed in organic solvents. When synthetic chemistry reaches its limitations, we can still rely on nature, as enzymes are known to have great selectivity control as well as sustainable reaction conditions. An enzyme family of great potential for oxyfunctionalizations is the Rieske non-heme iron dioxygenases (ROs). They can perform cis-dihydroxylation reactions on arenes and other substrates as well as allylic monohydroxylations in outstanding regio- and stereoselectivities. In nature, these enzymes are involved in the degradation pathways of aromatic compounds, but other substrates, like derivatized olefins, heterocycles and monoterpenes have been accepted as well. Aim of this study was to expand the substrate scope of ROs by the class of the sterically more demanding sesquiterpenes, as these molecules and their oxyfunctionalized pendants are of great industrial interest, for example as the sandalwood fragrance santalol, or bisabolol derivatives, as pheromones for plant protection. Additionally, the loop studies of Peter Heinemann should be expanded on, to develop his initial loop mutagenesis approach further and use molecular dynamics simulations to increase the understanding of these highly flexible and dynamic structures. In a first part, RO-WTs from the IBTB in-house library were screened with the sesquiterpene substrates β-bisabolene and α-farnesene, leading to the Cumene dioxygenase derived from Pseudomonas fluorescens IP01 as the only enzyme with starting activity towards the substrates. As the starting activity was residual, the expression conditions and the reaction system were optimized, leading to a set of optimized conditions and a 4-fold increase in product formation with β-bisabolene as screening substrate. Next up, the influence of loop insertions and deletions on the substrate specificity was investigated by applying the herein developed InDel scan, a screening method, where all amino acids of a loop are deleted separately, while also inserting an “Glycyl-prolyl” fragment in between each amino acid, leading to a condensed screening library. To test this approach, four different loops in the CDO were screened, a tunnel flanking loop, two helix-loop-helix loops and an indirectly interacting loop. This screening led to a strong shift in regioselectivity with β-bisabolene, leading from a ring-hydroxylation product 1a in the WT to 90% of a terminal-hydroxylation product 1b with just a single point deletion I288del. To identify the products by preparative biotransformations, various biotransformation conditions were tested, and the products were isolated with moderate yield (1a: 17 mg, 0.77 mmol, 38,5%; 1b: 24 mg, 0.11 mmol, 54.5%). And analyzed by NMR. To gain a deeper understanding of the underlying enzymatic mechanisms involved in this selectivity shift, molecular dynamics simulations were performed in cooperation with the Loschmidt Laboratories in Brno, Czech Republic. Molecular dynamics simulations with the free enzyme and with enzyme-ligand complexes were performed. For the latter, the ligand was docked into the active site and two conformations were chosen as starting conformation, one with the ring-carbon close to the catalytic iron in the active site (Ring-MD) and the other with the tail-carbon close to the catalytic iron (Tail-MD). The molecular dynamics simulations were evaluated and gave great XIX insight into the enzymatic properties. The B-factors of the free enzymes and enzyme-ligand complexes were calculated and showed that the variant I288del possessed a more rigid active site, which can potentially stabilize the ligand better. Additionally, CAVER analysis of the substrate access tunnel revealed that the tunnel is now open more often, suggesting that the substrate can enter the enzyme with less hindrance and also more frequent, leading to higher product formations. Analysis of distances between the catalytic iron and the reactive carbons in the substrate revealed that the reactive tailcarbon is positioned closer to the iron in the variant, leading to potentially more reactive conformations. The total interaction energies of the simulations were calculated, which showed that the ring-hydroxylation product is also energetically preferred in the WT, which is not the case in the variant anymore. In a final computational part, adaptive steered molecular dynamics simulations were performed, in which the ligand is positioned in front of the substrate access tunnel of the enzyme, and then pushed with a constant force into the active site of the enzyme. These simulations showed that, in the WT, the substrate preferably enters with the ring first, leading to a ring conformation in the active site. In the variant I288del this changes as the ligand is strongly preferred to enter the enzyme with the tail-carbon first, subsequently leading to tail-conformations in the enzyme and a stronger preference for the tailhydroxylation product. All these findings align with the experimental results, giving detailed insights into the ligand transport and tunnel modifications induced by the loop mutation. In a second experimental part, the goal was to push the product formation for the tail-hydroxylation product 1b by addressing the active site with site-saturation mutagenesis to fine tune the positioning of the substrate. This led to two new variants I288del_A321T (V1) and N279T_I288del_A321T (V2), with high selectivities and product formations of 1.35 mM and 1.58 mM, respectively, whereas the parent only converted 1.25 mM. Further rounds of site-saturation mutagenesis yielded no more increases in product formation or selectivity. To expand on sesquiterpene hydroxylation, santalene, bergamotene, germacrene D and valencene, which were initially unconvertible by the WT, were screened with the generated saturation variants I288del, V1 and V2. With these variants 3 of the 4 substrates could be converted, most remarkably the substrate santalene which yielded the industrially relevant santalol with a conversion of 0.74 mM. These findings led to a patent in cooperation with the company Isobionics (Geleen, NL). To expand on the InDel scan, the CDO InDel library was screened with two other substrates, (R)- limonene and 2-phenylpyridine, and searched for selectivity shifts, generalist variants or increased product formations. Variants, with either increased selectivities or activities, were found for both substrates. To expand this approach further, another enzyme, the naphthalene dioxygenase derived from Pseudomonas sp. NCIB 9816-4, was tested. Four active site- and tunnel flanking loops were chosen and mutated with the InDel scan, leading to a library of 56 variants for screening. For comparison, (R)-limonene and 2-phenylpyridine were chosen as screening substrates. The results were similar to the results of the screening with the CDO. A few variants with a completely changed product spectrum or high increases in product formation were found and for 2-phenylpyridine as substrate, and even a completely new, unidentified product appeared with the InDel variants. In summary, this work managed to convert sesquiterpenes with ROs in reasonable conversions, even achieving milligram scales for product identification. We achieved a complete selectivity shift in the hydroxylation of β-bisabolene from ring- to terminal-hydroxylation. The computational study gave great insight into the ligand transport in ROs, and the mechanisms conveyed by the loop mutation I288del, discovered by the InDel scan. The InDel scan was developed as an initial scanning tool to find start activity or modified selectivities and seems suited for ROs, also in later stages of a mutagenesis XX approach. It can also be used when active site mutagenesis yields no beneficial variants anymore and confirms the relevance of loop structures in enzymes once more.