We performed mRNA extraction on the Aspergillus nidulans strain transformed with Ti_OvaA, Ti_OvaB, Ti_OvaC, vrtD, and NphB and obtained the cDNA and performed a PCR of the cDNA and observed bands for both vrtD and NphB, the two genes we were investigating. We then purified the bands and sent them out for sequencing in the case that the genes may be mutated but the sequenced results displayed no mutations. With this information in mind, we decided to perform the NphB reaction with GPP and olivetolic acid using A. nidulans lysates expressing NphB. Zirpel et al. had demonstrated that whole cell bioconversion in Saccharomyces cerevisiae expressing wildtype NphB, olivetolic acid, and GPP did not produce CBGA; however, employing lysates of that same Saccharomyces cerevisiae strain in assays with supplemented GPP and olivetolic acid did produce CBGA as well the O-geranylated analog.Therefore we used A. nidulans lysates expressing NphB and supplemented olivetolic acid and GPP and we did observed CBGA production, much greater than what was observed in vivo. . We concluded therefore, that from the transcription and lysate data, that NphB was in fact correctly expressed in A. nidulans and that the issue could be low availability of GPP or that NphB is localized away from olivetolic acid and/or GPP. To further explore the issue of localization, we tagged the NphB enzyme C-terminal with green fluorescent protein using a flexible 5 amino acid linker . Microscopic images of the tagged NphB enzyme in A. nidulans displayed that the NphB was not localized in any punctuate organelles but rather was localized all throughout the fungal body indicating that the enzyme is located in the cytoplasm. 

As previously described,hydroponic flood table most of olivetolic acid and its analogs produced from our platform are found in the media indicating that the compounds are being secreted from the fungal body and therefore, olivetolic acid and its analogs go through the secretory channel in Aspergillus nidulans into the media. Therefore, we sought to localize the NphB to where GPP and the compounds were. We had to then understand where GPP was localized.Further genome mining in our lab for prenyltransferases harboring activity to produce CBGA from olivetolic acid and GPP, revealed a prenyltransferase similar to the ascA prenyltransferase from Acremonium egyptiacum, found in Colletotrichum higginsianum. Thisenzyme was discovered by Colin Johnson, a graduate student in the Tang Lab. The ascA prenyltransferase, a prenyltransferase belonging to the UbiA family, from Acremonium egyptiacum has been characterized to prenylate orsellinic acid with a farnesyl group.However, we demonstrated that the prenyltransferase from Colletotrichum higginsianum, labeled colA, was able to prenylated orsellinic acid with a geranyl group instead of a farnesyl group which is ideal since CBGA contains a geranyl group as opposed to a farnesyl group. Colin had searched through databases of isolated fungal products for C3 geranylated β-resorcylic moieties. He was able to find a couple compounds known as Colletorin B and Colletotrichum B that fit the description. Through genome mining, he was able to identify the cluster producing these compounds which contained a NRPKS, an NRPKS-like enzyme, a halogenase, and UbiA-like prenyltransferase which he labeled colA. Therefore, the colA gene was seen as a good candidate to test its ability to prenylate olivetolic acid and its analogs to CBGA and its analogs. Heterologous expression of colA with Ma_OvaA, Ma_OvaB, and Ma_OvaC as well as heterologous expression of colA with Ti_OvaA, Ti_OvaB, and Ti_OvaC unfortunately showed no production of prenylated olivetolic acid, unsaturated olivetolic acid, sphaerophorolcarboxylic acid, or prenylated unsaturated sphaerophorolcarboxylic acid; however, production of geranylated orsellinic acid was observed.

Further probing through the literature indicated that Aspergillus nidulans harbors an endogenous bio-synthetic pathway responsible for the production of orsellinic acid explaining the geranylated orsellinic acid result. Furthermore, the production of geranylated orsellinic acid did indicate that the GPP pool in Aspergillus nidulans is sufficient answering our concerns and therefore further indicating that localization is the key reason why NphB has not been shown effective in Aspergillus nidulans.ColA, a UbiA-prenyltransferase predicted to have seven transmembrane domains, could therefore not be purified and was subjected to feeding studies in both Saccharomyces cerevisiae and Aspergillus nidulans. Saccharomyces cerevisiae and Aspergillus nidulans strains expressing colA were supplemented individually with 200 µM orsellinic acid, 200 µM divarinic acid, 200 µM olivetolic acid and 200 µM sphaerophorolcarboxylic acid. The Saccharomyces cerevisiae feeding results demonstrated that colA was very efficiently able to prenylate orsellinic acid and to a lesser extent divarinic acid but although able to prenylate olivetolic acid and sphaerophorolcarboxylic acid, at very low efficiencies. A. nidulans feed results demonstrated that similar to S. cerevisae, colA was able to efficiently prenylate orsellinic acid: however, less so divarinic acid and prenylation of olivetolic acid and sphaerophorolcarboxylic acid was not observed. The differences between the feeding results of S. cerevisiae and A. nidulans were attributed to the fact that the S. cerevisiae strain was much more heavily engineered with regards to optimization of pathways than the A. nidulans strain and therefore was more optimal for secondary metabolite production. Prediction software indicated that the colA gene had its transmembrane domains localized in the endoplasmic reticulum . Additionally, regarding the mevalonate pathway responsible for the production of the intermediate GPP, one key enzyme in the pathway, hydroxymethylglutaryl-coenzyme A reductase , a rate determining enzyme responsible for the conversion of HMG-CoA to mevalonate163 was also predicted by TMHMM 2.0 to be located in the ER, which we hypothesize explained why colA was able to geranylate orsellinic acid and divarinic acid and why NphB, located in the cytoplasm, was unable to geranylate any B-resorcylic acid in vivo.

Faced with the difficulty that the engineered NphB which is able to efficiently geranylate olivetolic acid to CBGA is not able to do so in vivo based on localization issues and the issue that colA which is able to utilize GPP to prenylate orsellinic and divarinic acid in vivo but not longer alkyl chain variants, we decided therefore that fusion of the NphB enzyme C-terminal to colA would solve the localization problem, allowing NphB to utilize GPP to efficiently prenylate olivetolic acid and sphaerophorolcarboxylic acid. Once again, employing a flexible linker , we fused NphB C-terminal to colA and heterologously expressed the fusion product with both the Ma_OvaA, Ma_OvaB, Ma_OvaC and Ti_OvaA, Ti_OvaB, and Ti_OvaC set of genes in Aspergillus nidulans . LCMS traces of heterologous expression results did show that CBGA was in fact produced further indicating that localization was the key issue, but the CBGA production was at low levels which was perplexing. We purposed then to utilize this fusion approach with a wide variety of endogeneous A. nidulans enzymes localized in various membranes in the cell with the thought that there was possibly another region where the compounds and GPP were localized. We fused the NphB Cterminal to three other endoplasmic reticulum localized proteins: HMG-CoA reductase previously described, sec12p,ebb and flood table the guanine nucleotide exchange factor , specific for the SAR1 gene which acts as a regulator of COPII vesicle budding from ER exit sites , sec63p, encoding a protein essential for secretory protein translocation into the ER.In addition to localizing NphB to the ER, we also localized the enzyme to the peroxisome employing the peroxisome targeting signal 1 as well as to the nucleus using a nuclear localization signal . Not only did we localize NphB to the ER, peroxisome, and nucleus, we also fused the protein C-terminal to the mitochondria protein acetyl-CoA acyltransferase, responsible for converting 2 units of acetyl-CoA to CoA and acetoacetyl-CoA molecules.Lastly, we fused NphB to C-terminal to the plasma membrane protein tmpA, an oxidoreductase involved in the A. nidulans conidiation pathway.Similar to the colA-NphB construct, heterologous expression of all these tagged and fusion constructs were expressed in combination with Ti_OvaA, Ti_OvaB, and Ti_OvaC, the set of enzymes responsible for predominately producing olivetolic acid. LC-MS trace results showed that similar to the colA-NphB results, that in most of the fusion and localization tagged constructs, CBGA production was observed but at low levels. This could be due to the fact that NphB may not be folding as well in the fusion construct. Therefore, there is continued need to mine for other prenyltransferases.Blasting the colA enzyme across NCBI based genomes yielded a hit in the genome of Talaromyces islandicus having 55% identity to colA. Heterologous expression of this UbiA-type prenyltransferase from Talaromyces islandicus, labeled TislaUbiA, with the Ti_OvA, Ti_OvaB, and Ti_OvaC enzymes in Aspergillus nidulans showed the expected CBGA methyl variant result as well as prenylated olivetolic acid, albeit small, a result not observed with colA. Therefore, with four fungal genes, we were able to access those two cannabinoids, a result not observed before without utilizing genes from the Cannabis sativa plant. To increase the CBGA production utilizing the TislaUbiA enzyme, we would need to perform mutations in the active site of the enzyme responsible for binding to the aromatic prenyl acceptor. To do this, we had to identify the active site. TislaUbiA, colA, and CsPT4 are all UbiA prenyltransferases, membrane embedded prenyltransferases harboring two aspartate rich motifs associated for the divalent, cation-dependent prenylation.A crystal structure of archaeal UbiA in both its substrate bound and apo form was elucidated by Cheng et al. Cheng et al were able to obtain a 3.3 crystal structure of the archaeal organism Aeropyrum pernix UbiA The group observed that the structure of the enzyme contained nine transmembrane helices arrange counterclockwise with a large central cavity.

They also obtained a 360 crystal structure of ApUbiA in a substrate bound state, with the substrates p-hydroxybenzoic acid and geranyl thiolpyrophosphate activated with magnesium ions. In the crystal structure, GSPP was bound in the central cavity and a small basic pocket near the GSPP binding site was determined to be binding pocket for PHB binding. With this in mind, Cheng et al performed mutations to determine which amino acids were critical for binding. For the PHB binding pocket site, they determined that Arg43 and Asn50 were both critical to PHB binding.We therefore used this information to generate mutations for TislaUbiA with the purpose of opening the small basic pocket to accept large ß-resorcylic acid moieties. Using Alphafold, we generated a structural model for our TislaUbiA enzyme and comparted it to ApUbiA. The next steps, then would be to select for mutations that we postulate would open the binding pocket. Going back to Saccharomyces cerevisiae, we tested to see if we were able to achieve functional expression of CsPT4, the prenyltransferase from Cannabis sativa that Luo et al. characterized. We had decided to continue production of the cannabinoid bio-synthetic pathway in our model engineered A. nidulans host due to the high titers of olivetolic acid and its analogues that we were producing. We were unsure if changing the platform to S. cerevisiae would replicate the high titer production. Similar to Luo et al, we removed the N-terminal chloroplast targeting sequence of CsPT4. We heterologously expressed the enzyme in our S. cerevisiae super strain and subjected the transformed strain to feeding assays. We fed 200 µM of orsellinic acid and 200 µM of sphaerophorolcarboxylic acid and observed geranylation of both. We were able to produce the heptyl version of CBGA at moderate to high titer quantities, an exciting result since this is the direct precursor to THCP. We also saw that although we did take a hit in titer when we moved our platform to S. cerevisiae, we were still able to produce about 500 mg/L of SA. We then sought to achieve functional expression of THCAS. Production of the elaborated cannabinoids from CBGA involves the use of just one cyclase enzyme, with the final elaborated cannabinoid structure dependent on the cyclase enzyme employed. There are three elucidated dedicated cyclase enzymes from the Cannabis plant capable of cyclizing CBGA to the final cannabinoid: tetrahydrocannabinolic acid synthase , which forms tetrahydrocannabinolic acid from CBGA, cannabidiolic acid synthase , which forms cannabidiolic acid from CBGA, and cannabichromenic acid synthase which forms cannabichromenic acid from CBGA. All three of these oxidocyclase enzymes are part of the berberine-bridge enzyme -like family of enzymes, harboring a flavin adenine dinucleotide -binding domain, a substrate-bindingdomain, an N-terminal signal peptide, and a BBE-like C-terminus part of the FAD-binding module.