In Taxus, both taxadiene synthase and geranylgeranyl diphosphate synthase are localized to the plastids. This is important to note when introducing TXS in yeast, as localization sequence removal is key to improving performance, with truncation of the first 60aa yielding the largest taxadiene titers . Taxadiene-5α-hydroxylase then converts taxadiene to taxadiene-5α-ol , this is the primary modification of terpene scaffold. T5αol is then acetylated at the 5C position replacing the hydroxide. Taxadien-5α-acetate then undergoes numerous oxidations, an epoxidation, dehydration and benzylation to produce 10-deacetylbaccatin III. The final bioconversion of 10-deacetylbaccatin III to Taxol requires five additional enzymes: phenylalanine aminomutase, β-phenylalanoyl-CoA ligase, [baccatin III 3-amino, 13-phenylpropanyltransferase], taxane-2α-hydroxylase, and a N-benzoyl transferase . Currently, many of the enzymes required for Taxol biosynthesis have yet to be experimentally validated, though multiple research teams are working to characterize the pathway . The daunting task of engineering Taxol production into yeast is made even more challenging due to the spectrum of byproducts generated in the first two steps of the pathway, which drastically diminishes metabolic flux to downstream reactions. The taxadiene to T5αol conversion by T5αH is especially problematic in yeast, producing more than ten dead-end alternative products, with T5αol as a minor product. Prior to this work, the structure of most of these molecules were yet to be experimentally verified which complicates optimization of T5αOH activity and T5αol production. Tobacco, specifically Nicotiana benthamiana,indoor grow trays can serve as an excellent platform for investigating the biosynthetic characteristics of plant derived enzymes with Agrobacteriumbased transient expression.

Efforts to engineer plant natural product biosynthetic pathways in yeast can be greatly enhanced when coupled with these tobacco-based investigations. Especially for high-throughput gene discovery, as plant derived enzymes frequently require modification/truncation for proper functionality in yeast while remaining active in tobacco without modification. We sought to bolster our investigations into T5αH activity with parallel experiments in yeast and tobacco, with the goal of deciphering the T5aH product spectrum and confirming T5aol biosynthesis. Characterizing the array of oxidized products produced by the T5αH reaction with taxadiene is made more perplexing when compared to the GCMS analysis of Taxus cells grown in tissue culture. We found that the spectrum of oxidations generated by T5αH is a phenomenon only observed in our transient production systems , with these additional byproducts absent in Taxus tissue . This could suggest a secondary level of regulation in Taxus that restricts the “unproductive” oxidation of taxadiene at the 5C position, potentially via allosteric regulation of T5αH, the presence of peripheral proteins that modify reaction kinetics, or substrate shuttling via metabolon formation. Interestingly, while the product profile is fairly consistent in both the yeast and tobacco transient systems, the relative and absolute abundance of each molecule are quite diverse. Taxane production in yeast is carried out in a modified line engineered to overproduce GGPP , the primary substrate of the taxadiene diterpene scaffold. TXS was further modified to improve taxadiene production with the removal of the 60aa localization sequence, an N-term maltose binding protein fusion, and a C-Term ERG20 fusion . Two copies of the augmented TXS were integrated into the yeast genome with a galactose inducible promoter to allow for pathway activation after reaching desired cell culture density. After confirmation of taxadiene production, T5αH and a CPR from Taxus were integrated into the 2x TXS strain with a galactose inducible promoter for T5αol production. This strain was then used for metabolite extraction and GCMS analysis for comparison to the T5αH product profile generated in the tobacco system, as well as molecular structure confirmation.

More than eleven oxidized products, both mono-oxidized taxadiene and di-oxidized taxadiene , are produced in both the yeast and tobacco systems . This demonstrates that not only does the T5αH reaction result in varying oxidations of taxadiene, but also has the capacity for sequential oxidations of taxadiene, further complicating optimization of T5αol production specifically. Unfortunately, the ablation of non-specific T5αH activity will be crucial for the successful integration of the Taxol biosynthetic pathway in yeast. We have currently identified the structure of eight MOTDs and DOTDs produced by T5αH via product purification and NMR analysis. Four of these are MOTDs, one being T5αol and the other three forming epoxides from the hydroxide added at the 5C position . The four DOTDs identified in this work have a myriad of secondary oxidations, with one undergoing a rearrangement of the primary taxadiene scaffold . These data, in conjunction with findings that these alternative products are not found in Taxus cells, suggest that the T5αH reaction with taxadiene is not specific in these transient hosts, with the capacity to form numerous transition states during the reaction. This could potentially be attributed to a sub optimal redox coupling between T5αH and the CPR257 . Even though both enzymes were mined from Taxus, the interaction between them may be altered when outside their native context or orientation. This also strengthens the argument that additional, unknown/unidentified structural proteins or redox intermediaries are present in Taxus that modulate the reaction. One difference observed when comparing the GCMS data from tobacco and yeast is the relative abundance ofOCT compared to T5αol . In yeast, the OCT/T5αol ratio is ~2/1, while in tobacco OCT and T5αol levels are relatively equal . Overall, the differences observed are most likely attributed to altered enzyme behavior in yeast and tobacco, as interactions with endogenous proteins and substrates would be quite varied in the plant and fungal host. Comparisons between the absolute abundance and production titers for these products in both systems are not currently viable, as there was no standardization done to normalize between samples. Another important aspect of P450 activity that needs to be explored in future experiments is the balancing of CPR and P450 levels. Over expression of both the CPR and P450 can lead to a suboptimal redox coupling if the stoichiometry of CPR and P450 are not properly balanced. Experiments that alter the level of CPR expression with constitutive promoters of varying strength could inform decisions for optimizing the concentration of redox partners, which in turn should improve T5αol production in yeast.

A major challenge when engineering functional P450s from plants into yeast is maintaining robust expression and proper ER localization. P450s, along with their general redox partners CPR and Cytochrome b5 reductase, require ER anchoring for proper assembly and functionality of the electron transport chain coupled to the oxidative reaction carried out by the P450. Confocal microscopy is a great tool for the examination of enzyme expression and localization dynamics when coupled with fluorescent reporter protein fusions* . To better understand how enzymes from plants behave in yeast, we generated a suite of fusion proteins consisting of an enzyme of interest and a fluorescent protein for visualization of expression/localization dynamics . In addition to our investigation of specific enzyme behavior in yeast, we sought to better understand how the expression vector used in each experiment influences protein production across a population. A common method for enzyme characterization in yeast utilizes high-copy plasmids with an auxotrophic selection marker for expression*. Unlike most antibiotic-based selection methods, kanamycin selection in E. coli for example, auxotrophic selection has a degree of inefficiency. It is possible for “cheaters” not carrying the plasmid containing the selectable marker to persist by scavenging the deficient amino acid from dead cells in culture. When comparing the expression patterns of an integrated fluorescent protein and various fusion proteins carried on an auxotrophic plasmid we observed a drastic effect on expression across the population . Using either of the plasmid-based vectors for fusion protein expression resulted in a mixed population when visualized, specifically when analyzing the binary pattern of expression vs. no expression in each cell . While other factors likely contribute to this phenomenon,vertical grow racks for sale there is a consistent pattern observed when comparing the expression of stably integrated and plasmid-based expression vectors. Even though genetic integration of an expression vector yields homogenous expression in a population, and thus consistency in production titers and peripheral observations, plasmid-based expression has many utilities when engineering biosynthetic pathways in yeast. While not advised for strains engineered for large-scale production, plasmid-based expression is excellent for initial experiments aimed to characterize unknown biosynthetic enzymes. This is due to the high throughput manner in which enzymes can be tested and optimized prior to genetic integration, providing a primary system for investigating novel biosynthetic pathways in yeast while building a final production strain. A key observation can be made when comparing the localization pattern of a full-length dioxygenase with that of the same enzyme with the localization sequence truncated. Expression of the full length sequence in yeast results in a unfunctional enzyme, which is corroborated by the abnormal localization pattern. Truncation of the native localization sequence restores biosynthetic functionality as well as the proper cytosolic localization in yeast . These data demonstrate the utility of confocal microscopy for the visualization of enzyme behavior, which can inform the interpretation of production titers when rebuilding biosynthetic pathways in yeast. Another aspect of enzyme expression and localization we wanted to explore was stability over time. Taxadiene-5α-acetyltransferase , a soluble and cytosolic enzyme from the Taxol pathway that acetylates T5αol, was fused to eGFP and visualized over time post-induction . At 8hr post-induction TAT:eGFP is seen properly localized to the cytosol with minor puncta forming in a few cells. Though, when visualized at 24hr PI the majority has been shunted into numerous puncta in each cell, with limited fluorescence remaining in the cytosol. Complete amalgamation of TAT:eGFP into these puncta can be seen at 48hr PI. This could be an explanation for the lack of biosynthetic activity for TAT when introduced into the T5αH production strain, though minimal levels of T5αol could also restrict activity. While the mechanism responsible for this shift from dispersed cytosolic localization to compartmentalized bodies has not been identified, it is hypothesized that the enzyme is being trafficked into proteosomes or peroxisomes for turnover.

Especially when compared to a peroxisome localized fusion protein, which shows a similar pattern seen at 48hr PI . Observing this localization pattern over time has highlighted additional aspects for optimization when engineering biosynthetic pathways in yeast, persistence of localization and stability/half-life. Ensuring that the enzymes required for target molecule biosynthesis have robust expression, along with persistence in both localization and half-life, is crucial when optimizing production strains for large scale synthesis schemes. An additional characteristic of enzyme behavior in yeast was observed when comparing localization dynamics over time of an ER and a cytosolically localized enzyme. Ketoreductase 11 and 23 involved in saponin biosynthesis were selected for visual comparison. KR11 was tagged at the C-term with eYFP and KR23 was tagged at the C-term with CFP to analysis localization simultaneously. KR11:eYFP and KR23:CFP were cloned into the same plasmid-based expression vector with Gal1 and Gal10 driving their expression. When visualizing the localization of both KR11:eYFP and KR23:CFP at 8hr PI, KR11:eYFP is seen properly localized to the cytosol while KR23:CFP is properly localized to the ER membrane . After confirming both enzymes were expressed and properly localized at 8hr PI, we visualized the same culture at 24hr PI. As previously observed with other cytosolically localized enzymes, KR11:eYFP had been shunted into small bodies with very little remaining in the cytosol. KR23:CFP on the other hand retains proper ER localization in most cells, while showing limited loss in overall abundance . This phenomenon demonstrates an incongruity in behavior between ER-anchored and cytosolically localized enzymes from plants when expressed in yeast. The ER localized enzymes are largely occluded from accumulation to the “unknown bodies” that capture the cytosolic enzyme. While the mechanism of action responsible for these observations has yet to be characterized, these findings highlight aspects of enzyme behavior that should be explored when engineering biosynthetic pathways from plants into yeast. Unfortunately, the P450s required for Taxol biosynthesis fail to properly localize to the ER membrane when expressed in yeast. This issue presents a major challenge for engineering a functional Taxol biosynthesis pathway due to the required redox coupling with ER bound enzymes. One option for reconstituting proper ER localization is to exchange the native ER anchor region of the P450 with an anchor sequence known to properly bind the ER membrane in yeast. This chimeric protein would hopefully have restored ER localization while retaining the native biosynthetic characteristics.