Biosurfactants, surfactant molecules originating from microbial synthesis and/or renewable resources, have been proven to be a valuable alternative to the (petro)chemical based variants widely used today. Due to the structural variability of these molecules, a wide variety of (potential) applications ranging from household usage to pharmaceuticals have been identified. Thanks to their biological degradability and production using renewable resources combined with their (biological) activity even at low concentrations, the focus of surfactant production is slowly turning towards these molecules. Still, widespread usage of these fascinating molecules remains low mostly due to the higher production cost and limited structural variability of the commercialised compounds. When looking at glycolipids, carbohydrate or lipid variety is often limited to just a handful of different types of groups with minor modifications. In the case of sophorolipids, the fatty acid tail is most often limited to a C18 fatty acid, either with its carboxylic group free or condensed to the sophorose moiety. Longer or shorter congeners are possible, but only in low amounts or as a part of a more complex mixture of molecules.

When taking productivities into account, only a few microorganisms are capable of producing enough biosurfactant in a short amount of time to reduce the cost of making them. Unfortunately, one has to take into account certain drawbacks like potential pathogenicity of the producer strain. Using robust producer strains for the production of wild type, modified or new-to-nature molecules has been proven to be an interesting technique to enhance the production of these molecules. Both on a genetic and process level optimisation has been done with varying degrees of success.

During this thesis, S. bombicola was turned into a robust strain for the production of different kinds of molecules. Both genetic engineering and better understanding the production parameters allowed to gain deeper insights in how S. bombicola produces biosurfactants and how to steer the production towards novel kinds of molecules.

In the second chapter, the main focus was creating new molecular tools for easier modification and/or screening of novel strains. A first tool was the development of an episomal vector for S. bombicola. Though many yeast species have one or several plasmids, no such thing is known for S. bombicola. By using in silico characterisation techniques and using heterologous and homologous sequences in a suitable screenings vector, the aim was to identify ARS sequences. Unfortunately, no active sequence could be found. The second tool is the maximum size of the integration cassettes used for knock-in and -out strategies. As it became clear that no clear upper limit could be found up until 11600 base pairs, a more practical was followed to double the genetic cluster of sophorolipid production. Though the idea was that more expression would result in more production, the results were not straight forward pointing towards higher transcription, translation and final productivity. Still, it proved that large constructs can be designed and transformed efficiently in S. bombicola, opening the possibility of introducing complex multistep pathways in a few steps. The final tool can be seen as multiple smaller tools together. Firstly, new kinds of fluorescent proteins were tested successfully in S. bombicola besides the already available GFP. Secondly, it was shown that coupling multiple domains does not necessarily impair protein function. This opens the possibility of creating larger chimeric enzymes where the catalytic domains are coupled for enhanced turnover.  

Thirdly, using the sophorolipid biosynthetic enzymes for the coupling resulted in their cellular localisation and provided a tool for comparing productivity with expression levels of the enzymes involved. When integrating this knowledge with qPCR data, better fine tuning of gene expression levels can be achieved.

Chapter three and four both focus on modifying either the carbohydrate or the lipidic group. In chapter three, is was proven that introducing novel P450s in S. bombicola can be an interesting strategy to alter the fatty acids being incorporated. Several self-sufficient P450 were introduced in S. bombicola for the production of new kinds of sophorolipids. Their activity could be measured, but no new molecules were produced. Engineering of the self-sufficient P450s by altering the critical phenylalanine at position 87 did not result in more favourable intermediates being produced. Still, they can be used as a starting point for other strategies. One of those was explored in depth by creating self-sufficient chimeric variants of CYP52A4 and CYP1. Though the production levels remained low, a high product uniformity was achieved for the CYP1BMR strain. Further optimisation including the substrate used resulted the production of C16 sophorolipids, both acidic and lactonic.

In chapter four, the main concept was modification of the carbohydrate group. The production levels of glucolipids and cellobiose lipids produced by previously designed strains remained low. Redesigning the original constructs as well as using optimized engineering strains resulted in the production of glucolipids up to 130 g/L. Interestingly, new molecules were observed as well in the produced mixture. It was already known that the UGTA1 transferase is capable of coupling a glucose moiety to both hydroxyl and carboxyl groups, the observation of bola glucolipids remained elusive until the first growth trials with the newly engineered glucolipid strains. For the cellobiose lipids, higher product uniformity was achieved by only expressing the ugt1 gene from U. maydis in S. bombicola. This not only resulted in losing the contaminating glucolipid fraction from the older strains, it also showed that UGT1 is solely responsible for the sequential glucosylation of the hydroxy palmitic acid. Further optimisation of the production strains is necessary to produce these molecules in bigger volumes, but for the first time relatively uniform cellobiose lipids have been produced in a non-wild type producer.

Chapter five was about turning S. bombicola into a producer of molecules that are unknown to this yeast. Rhamnolipids are molecules that originate from a bacterial producer and are structurally not similar to sophorolipids or cellobiose lipids. Production of these molecules involved integrating several bacterial pathways for the production of the necessary precursors. To obtain the hydrophobic part, the genes necessary for the bacterial FAS II system were introduced at two different locations in the genome. As a proof-of-concept, the production of HAA molecules was attempted. After several growth trials, it became clear to no activity could be measured. For the production of activated rhamnose, two different pathways were designed resulting in dTDP-L-rhamnose or UDP-L-rhamnose. To assess the availability of these rhamnose donors, a proof-of-concept was designed around the rhamnosylation of quercetin by the transferase At78D1 from Arabidopsis thaliana. After a growth trial, it became clear that only the strain producing dTDP-L-rhamnose was capable of producing a rhamnosylated product. A final experiment was carried out by combining cell lysates of both the NDP-strains equipped with the second rhamnosyltransferase RhlB and the HAA-strain. No production of rhamnolipids or related molecules could be observed.  

In conclusion, utilizing the robustness of S. bombicola for creating a platform for biosurfactant production is a valuable strategy. Though several obstacles remain concerning the fundamental knowledge of the organism, it is already possible to engineer strains that are capable of producing novel molecules on scales relevant for industrial production. The results in this work show that S. bombicola is a relevant player in the field of glycolipid production.