Genome engineering of filamentous fungi for efficient novel molecule production

The use of fungi by humans dates back more than 5000 years. At first a trial-and-error approach was used. Later the molecules in extracts became known and characterized. Now, with more and more species having their genome sequenced, effort in employing fungi focus on a genetic level, identifying new enzyme functionalities and engineering genomes for increased yield or novel product discovery. In this thesis gene engineering as well as gene discovery is applied to obtain novel enzyme functionality and molecules. The most important methods used are described in Chapter 2, while additional methods will be described in the respective chapters. ❧ Domain swapping in Non-Ribosomal Peptide Synthetase-like enzymes (Chapters 3 and 4) ❧ A group of small non-ribosomal peptide synthetase (NRPS)-like enzymes was recently discovered in Aspergillus terreus. The products of these enzymes include antitumor metabolites asterriquinones and butyrolactones, and the phytotoxic phenguignardic acid. Knowledge of the mechanism of these biosynthetic enzymes will open the way for developing analogues with improved functionality. NRPSs are typically megadalton size enzymes that produce peptides by selecting free amino acids, activating and tethering them, and via a condensation reaction couple one to the next. Each amino acid is recognized by a protein domain so a particular NRPS makes a specific peptide sequence, unlike the ribosome which uses the DNA as template. In non-ribosomal peptide formation, non-proteinogenic amino acids can be incorporated and via additional protein modules the peptide can be modified during chain elongation, for example epimerization to generate D-amino acids, N-methylation, or side chain cyclization to generate oxazolidines. The immunosuppressant cyclosporin is the most famous product of this type of enzymes. Their modular architecture, with a separate domain for each amino acid and reaction, makes them a target of engineering efforts. Libraries of NRPs could be generated by recombining the domains that recognize amino acids. To date much effort has gone into determining the domain boundaries and where exactly to exchange protein domains while retaining whole protein functionality. The main downside of the engineered enzymes has been yields as low as 1% of the wildtype enzyme. To combat potential low yields for this project, a heterologous expression system was used: Aspergillus nidulans and its powerful native alcA promoter. NRPS-like enzymes are similar in domain architecture to their NRPS cousins. The main difference is that they are a lot smaller with typically only three domains: one for recognizing and activating an amino acid (A domain), one for tethering (T domain) and one for release (TE domain). The other difference is that instead of an amino acid, an alpha-keto acid is activated, which means the final product does not contain any peptide bonds. Five enzymes with these three domains were recently identified in A. terreus and their products were confirmed via knockout studies and heterologous expression in A. nidulans with yields of 50-100 mg/L. The products were all dimeric aromatic alpha-keto acids, differing only in cyclization pattern. Our hypothesis was that the cyclization pattern is determined by the TE domain so cyclization and release go hand in hand. To test this, we generated hybrids of the genes by fusing the TE domain of one gene with the A and T domain of another. The resulting functional hybrid enzymes had comparable yield and allowed for the structural characterization of their products, showing that the cyclization pattern of hybrids was determined by which TE domain they contained. Products of these hybrids were novel secondary metabolites with new combinations of heterocycles and side chains. This result shows that our system can not only be used to study enzyme mechanisms, but also engineered to produce new molecules on an industrial viable scale. To generate even more novel molecules, the tailoring enzymes of the wildtype clusters were coexpressed. Using this approach specifically methylated and prenylated molecules were successfully obtained. Part of this work was published in Organic Letters in 2016. Future work will focus on generating more functional hybrids and thus more novel molecules. These new molecules will be compared to the wild type compounds in their phytotoxic (plant assay using leaf discs) and cytotoxic activity (MTT assay) in four cancer cell lines, NCI-H460 (non-small cell lung), MCF-7 (breast), SF-268 (CNS glioma), and MIA Pa Ca-2 (pancreatic). ❧ Heterologous expression of Ribosomally synthesized and Post-translationally modified Peptides (RiPPs) (Chapter 5) ❧ Ustiloxin B is a cyclic tetrameric peptide exhibiting antimitotic properties via microtubule inhibition making it a cancer drug lead. It is produced by a number of crop infesting fungi like Aspergillus flavus and Ustilaginoidea virens and until recently it was thought to originate from an NRPS. However, the NRPS was never identified and through genome analysis a gene with a 16 repeat of the four amino acids corresponding to the ustiloxin B sequence was found and shown to be a ribosomally synthesized precursor for ustiloxin B. Through knockout studies the gene cluster borders were determined, and there are 16 genes in the cluster. The generation of ustiloxin B analogues could lead to stronger antitumor activity and generally better drug properties. ❧ NRP analogues can be created through A domain replacements in NRPSs, but in the case of ustiloxin B a simple codon change in the precursor gene can yield analogues. To achieve this, we first expressed the gene cluster in our heterologous host A. nidulans, which has a ready genetic system with three usable selection markers. This way modifications to the ustiloxin B cluster can be made easily. Previously biosynthetic gene clusters were expressed one gene after the other, starting with the core gene and adding tailoring enzymes later recycling a marker with each gene. However, for a 16 gene cluster this would take 16 rounds of transformation. Therefore, a new method was developed for transformation of the complete gene cluster of A. flavus into the A. nidulans host in a single step, reducing the time needed from months to weeks. Culturing of the mutants under standard glucose minimal media (GMM) did not yield ustiloxin B. However, it is known that ustiloxin B is naturally produced by A. flavus when growing on and contaminating crops. Therefore, the mutants were grown in V8 juice media, in which amounts of ustiloxin B were detected. Subsequent overexpression of the transcription factor of the cluster, using the alcA promoter, increased ustiloxin B production tenfold. Future work will involve the expression of modified precursor peptides to generate ustiloxin B analogs. These will be tested for their antimitotic properties using initial microtubule assembly assay, followed by MTT assay to test the cytotoxicity on MKN-1, MKN-7, MKN-74, RERF-LC-MA, SBC-5, MCF-7, WiDr, SW-480, and KU-2 cancer cell lines. ❧ In this work it is shown that besides the enzymes Nature provides us, genome engineering can open up a chemical space currently not occupied, with seemingly endless possibilities. As engineering efforts increase, guided by structural data and high-throughput screening, one day a new natural enzyme will be discovered that has the same functionality as an existing engineered one. Until that day efforts on engineering and natural discovery both have value.