2023 Recap

In 2023, the USP-EEL-Brazil iGEM team decided to develop a project aimed at producing cannabidiol, a phytocannabinoid originating from the secondary metabolism of the Cannabis sativa plant, using the yeast Saccharomyces cerevisiae as a chassis. To carry out this proposal, in that year we focused on developing the project using mathematical modeling tools to consolidate our proof of concept and developing our parts and wet-lab strategy, so in 2024, we could return with the CBDynamics project with the wet lab experiments to complement it.

In terms of the methodology developed in 2023, the CBDynamics project focused on two primary metabolic pathways: the mevalonate pathway, which produces GPP, and the olivetolic acid pathway (Zirpel et al., 2017). Saccharomyces cerevisiae was chosen as the chassis because, among other reasons, it naturally synthesizes GPP, requiring only the introduction of the olivetolic acid pathway to produce CBDA.

To ensure the effective synthesis of the five enzymes in the biological circuit, we applied T2A sequences into our biological circuits, allowing the production of multiple individual proteins from a single mRNA molecule. This method exploits the eukaryotic ribosome's inability to form a peptide bond between the final proline and glycine residues, thus creating separate proteins without interrupting synthesis (Mansouri, 2014).

The methodology also included structural modeling of the five enzymes, phylogenetic analyses, molecular docking, and expression evaluations. These analyses guided the development and optimization of the biological circuits for the chosen chassis, ensuring a robust proof of concept. For detail information, see https://2023.igem.wiki/usp-eel-brazil/model

STRUCTURAL MODELING

To develop our strategy, we began by searching for all the sequences of the five enzymes in databases such as NCBI. Obtaining these FASTA sequences, structural modeling by homology was carried out using the SWISS-MODEL software for the five native enzymes (Figure 1).

In which, A represents the CsHCS1 model; B, the dimeric OS; C, the dimeric OAC; D, the CsPT4 and E, the CBDAS model.

After obtaining the initial models, structural modeling was performed again, this time incorporating T2A sequences using PyMOL's MODELLER extension. This allowed the addition of these new amino acids at the beginning, end, or both ends of the original protein sequence. To assess the stability of the modified models, ProCheck and MolProbity were used, identifying the most stable and energetically favorable structures. These evaluations enabled the team to observe structural differences due to the T2A sequences and select the optimal constructions for assembling the necessary biological circuits.

Expression vectors were constructed using the plasmids pRS425 and pRS426 using the Gibson Assembly method, a technique used to clone DNA fragments continuously. Around 72 biological circuits compatible with the expression system were designed from Snapgene, using the enzymes along with the T2A sequences.

Finally, we conducted a phylogenetic analysis by running MUSCLE-aligned sequences on the Jalview software, in which we compared the sequences of the enzymes in Cannabis sativa with other sequences of similar proteins in different families and genera. This allowed us to study whether the protein domains had been maintained, confirming that the catalytic sites had not been altered and that there were no problems related to the structural integrity and function of the enzymes. This analysis also allowed us to discover uncatalogued characteristics of the enzymes studied, such as some binding and catalytic positions.

MODELING FOR MOLECULAR DOCKING

To complement our analyses, we performed molecular docking to model interactions between proteins and substrates, focusing on conserved residues to understand how different substrates interact with our five proteins of interest, to gain insights into how the protein works and how its activity may have evolved to play different biological roles.

For validation, phylogenetics helped identify catalytic and binding sites, while the CavityPlus server (Wang et al., 2023) confirmed potential catalytic pockets. The top docking results from the HDOCK server (Yan et al., 2020) were compared with existing data, ensuring the best enzyme-substrate models for the metabolic pathway.

In this figure, “A” represents the CsHCS docked with Hexanoic Acid; “B”, “C”, “D” and “E”, the OS docked with different substrates - respectively, Hexanoyl-CoA and Malonyl-CoA in three conformations -; “F”, the OAC docked with Olivetol; “G” and “H”, CsPT4 sequentially docked with Olivetolic Acid and GPP, and finally, CBDAS docked with CBGA and catalyzing the reaction to produces CBDA.

EXPRESSION MODELING

To model the expression and production of CBDA (cannabidiolic acid) and its subsequent dehydration to CBD (cannabidiol), our team utilized the Mixed Effects (ME) model, a statistical approach representing gene expression after transformation (Llamosi et al., 2016).

The ME model accounts for the relationship between transcriptional and translational parameters, reflecting how these factors influence gene expression. This model provides insights into the cellular processes and helps identify bottlenecks in the modified metabolic pathways. By applying regression analysis, we determine key constants, enhancing our understanding of CBDA production.

Additionally, we employed logic gates from synthetic biology, similar to electrical engineering, to conceptualize the genetic circuits as a productive system. At the interface between electrical engineering and genetic engineering, doors emerged as an analogy for visualizing and understanding the cell as a productive process, a chassis.

Thus, thinking about Boolean logic for biomolecules, as described by Miyamoto et al. (2013), we identified some determinants as our gates: a, being the CDS promoter that first encodes all other enzymes of the biological pathway, in addition to the subsequent proteins, since they are interdependent on each other.

ACHIEVING EXCELLENCE THROUGH SYNTHETIC BIOLOGY AND COMMUNITY ENGAGEMENT

Recognizing the importance of public outreach, our team conducted data collection in Lorena's central square to gather insights on public perceptions of CBD and raise awareness of our research. These interactions revealed widespread misinformation and stigma surrounding cannabidiol, often rooted in a lack of knowledge. Educating the public is crucial to overcoming these barriers and promoting the medicinal benefits of cannabis-derived treatments.

To further our mission, we organized educational events, discussion groups, and contributed to the Brazilian Synthetic Biology Olympiad (OBBS), fostering broader engagement with synthetic biology and encouraging the next generation of scientists. Through these efforts, we aim to expand awareness of synthetic biology and its potential to improve lives.

We know that synthetic biology is a rapidly advancing field, applying engineering principles to biological challenges, offering powerful tools to address global issues, such as environmental crises and drug development. In our previous wiki , you can see how the 2023 CBDynamics project exemplifies the use of bioinformatics to drive excellence in synthetic biology. However, addressing real-world problems also requires direct engagement with affected communities. When we are dealing with people trying to solve real problems, dealing directly with the affected community is an indispensable part of the project. Seeking effective solutions using only the support of academia can be extremely ineffective. Therefore, during the development of the CBDynamics project, we sought to break down walls and have close contact with the population.

To engage directly with the local community, our team conducted, in 2023, surveys in the central square of Lorena, the city where our university campus is located. These surveys were carried out at different times and days to reach a broad audience. In addition to gathering public opinions on CBD, we used this opportunity to share information about our project and research. This outreach revealed a significant lack of knowledge regarding CBD and highlighted that most prejudice surrounding cannabidiol stems from misinformation, rather than inherent negativity associated with the compound.

Despite its medicinal properties, cannabis remains heavily stigmatized. Many people, often due to lack of awareness, do not associate the plant with its therapeutic potential. Correct public guidance is essential to dispel this stigma and ensure that individuals with conditions treatable by THC-based therapies feel comfortable seeking and following through with their treatments.

A key goal of our team, along with the Synthetic Biology Club, is to promote awareness of synthetic biology. Throughout the course of our project, we organized educational workshops and discussion forums to disseminate information and address any misconceptions through open dialogue.

Moreover, our project contributed to the Brazilian Synthetic Biology Olympiad (OBBS), a competition aimed at promoting synthetic biology and inspiring young students to engage with this emerging science. This initiative is part of our broader commitment to advancing scientific knowledge and outreach.

Modeling

INTRODUCTION

When we search for excellence, we should firstly go back to the initial concept of what we try to excel at. In the case of synthetic biology, we seek for ways of redesigning systems already found in nature through new biological systems, aiming for useful, ethical results to society as a whole. We are able to do that through the application of molecular biology tools and techniques associated with microbiology, combining them with engineering methods (Cameron; Bashor; Collins, 2014).

With that in mind, our team chose several different approaches to achieve our main goal - an optimized, scalable, market-competitive synthetic production of cannabidiol (CBD), our bioproduct of interest. As shown in extensive detail in our 2023 wiki, Last year our main goal was to introduce one metabolic pathway into our chassis, Saccharomyces cerevisiae, in order to make it possible for it to produce CBD. To carry out this proposal, we focused on developing the 2023 project using mathematical modeling tools to consolidate our proof of concept and developing our parts and wet-lab strategy - which we put to good use this year.

However, as a team of to-be engineers, it seemed natural to think how exactly we could make our synthetic CBD competitive in the market against its original counterpart and, most importantly, could we make it to be economically viable? This thought process was what led us through all of our modeling projects this year, and is a vital part of how we put engineering concepts as our guide to conceptualize our project as whole.

OPTIMIZATION CASSETTES

This year, optimizing CBD production was our primary objective, especially considering the fact that, by improving the ways of our biomanufacturing, we could create a process that is both optimized and competitive in future market insertions. Considering our goal, therefore, our modeling team was the main tool to promote our goals and reach the final results we expected. For that, our first idea was to diminish the expression of a gene that deviates the GPP to a different pathway, since that considerably affected our ending results, given the fact that, in our understanding, GPP is a predecessor of our metabolic way, naturally produced by S. cerevisiae and an intermediate metabolite through which our CBDA production line by the 5 enzymes previously cited will be induced. Therefore, by increasing the build-up of GPP in the cell, we believe we’ll have more of this compound turned into enzymes and, consequently, more CBDA, given its part on the metabolic pathway we used to understand its production.

To cause this build-up, we researched and found two main approaches: to block the deviation of GPP to other pathways other than ours or to super express an enzyme to synthesize GPP in the cell. For the first approach, we used the RNAi - which is the interference RNA that consists in a strand that will be produced but not read by the ribosomes and, in turn, not generate the protein transcribed there - and the RISC system, a group of enzymes that are responsible for the identification and cutting of the iRNA, without which our process would be unable to operate. In this situation, we generated two cassettes: the iRNA and the RISC system. The introduction of the RISC system in our cell is due to the fact that S. cerevisiae does not produce it naturally and without it the iRNA would not be functional (Pratt & MacRae, 2009). Our second idea, to super express GPP, was constructed based on the idea of the enzyme GPPS that converts IPP into GPP and, for that, we created one cassette (Luo et al., 2019).

Additionally, all cassettes fabricated in 2024 were anchored by homologous regions - segments of the yeast genome strategically inserted into the 5' and 3' untranslated regions (UTRs) of the cassettes. These regions would serve as signposts, guiding the yeast's cellular machinery to integrate the cassettes into its genome. To ensure that this integration didn't disrupt the yeast's vital functions, we meticulously searched for regions within the yeast genome that could be silenced without compromising its overall health. Our investigations led us to utilize knockout protein sites, each uniquely tailored for the specific cassette (Chia et al., 2018).

Finally, to facilitate future modifications, we strategically incorporated polylinkers - molecular beacons that guide enzymes to specific locations - into both the iRNA and SuperExpression cassettes (Amen & Kaganovich, 2017). These polylinkers were carefully positioned using zero-cutter enzymes, ensuring the integrity of our genetic constructs. This approach safeguarded the critical regions of the 5'UTR, 3'UTR, and the coding sequences of the ERG20 and GPPS genes, allowing for precise and targeted alterations without compromising their functionality.

2. KINETIC MODEL

To enhance the industrial production of cannabidiol (CBD), the USP-EEL-BRASIL 2024 team concentrated on optimizing enzymatic reactions within the CBD biosynthesis pathway. Understanding these reactions is vital for minimizing time and reagent waste, thereby facilitating large-scale execution. By employing kinetic modeling, the team sought to gain insights into each enzyme's activity profile, particularly how variations in substrate and product concentrations affect enzyme efficiency and reduce inhibition.

A central focus of this investigation was the calculation of the ideal theoretical concentration of hexanoic acid, a key substrate for maximizing the production of cannabidiolic acid (CBDA). This modeling approach, especially through the Michaelis-Menten equation, provided insights into reaction dynamics and potential bottlenecks within the biosynthetic pathway (Leow & Chan, 2019). By determining essential kinetic parameters such as maximum reaction rates (Vmax) and Michaelis constants (Km), the team could predict enzyme functionality under various conditions and refine their strategies for subsequent wet lab testing.

The enzymes studied included cannabidiolic acid synthase (CBDAS), olivetolic acid cyclase (OAC), prenyltransferase 4 (CsaPT4), hexanoyl-CoA synthase (CsHCS1), and olivetol synthase (OLS). The team collected kinetic constants from reputable sources, including the Brenda Enzyme and KEGG databases, to inform their models and elucidate how specific concentrations could enhance enzyme activity (Wiles et al., 2022; Stout et al., 2022).

The findings highlight the significant role of kinetic modeling in understanding enzyme behavior and identifying critical parameters that influence CBDA production. Although challenges arose in obtaining complete kinetic data for some enzymes, such as OAC, the team effectively utilized bioinformatics tools and phylogenetic analyses to estimate missing values for CsaPT4. This modeling effort not only optimized enzyme activity conditions but also revealed the intricate relationships between enzyme concentrations and their substrates, paving the way for more efficient CBD synthesis in future experiments.

To address the uncertainties arising from the search for kinetic data pertaining to the OAC enzyme, a comprehensive data review was conducted.

This review established that OAC belongs to the DABB family of enzymes, specifically classified as α+β barrel polyketide cyclases. Consequently, to mitigate the challenges associated with the calculation of kinetic parameters for OAC, it was determined that the kinetic parameters of OLS could serve as a valid indirect approximation for the behavior of OAC (Yang et al., 2016; Gagne et al., 2012).

_{Figure 6: CBDAS enzyme kinetic velocity profile graph Source:Authors (2024)}

_{Figure 7: CsaPT4 enzyme kinetic velocity profile graph Source:Authors (2024)}

_{Figure 8: OLS enzyme kinetic velocity profile graph Source:Authors (2024)}

_{Figure 9: CsHCS1 enzyme kinetic velocity profile graph Source:Authors (2024)}

The team's work resulted in the estimation of Km and Vmax values for OLS, addressing gaps in the literature. They also produced four graphs depicting the relationship between substrate concentration and reaction time, which illustrated the kinetic velocity profiles of each enzyme. These visualizations deepened the understanding of enzymatic mechanisms and facilitated the determination of a theoretical hexanoic acid concentration, assuming that all other reagents were maintained in excess and that conversion from reagents to products occurred without significant losses. This theoretical value provides a foundation for dry laboratory testing, thereby improving the efficiency of CBDA production.

In summary, this research demonstrates the vital role of kinetic modeling and computer simulations in elucidating enzyme activity, ultimately guiding best practices for biotechnological applications in CBD production. The systematic approach established here lays the groundwork for future experiments aimed at increasing the efficiency of biosynthetic pathways in industrial contexts.

3. BIOREACTOR

The concept of doing a bioreactor’s modeling for this project came from the idea of producing a large amount of biomass that we could, then, induce it to produce CBD. The reasoning is simple: the more we increase the number of viable yeast cells, the more microorganisms we have to induce production of CBD in their stationary phase - which we are able to do because the biological circuits we developed last year and then introduced into our yeast cells have inducible promoters. This reasoning gives us full control of the process, making it possible to model scale-up simulations for our project.

And how do we do that? Basically, our goal was to create an optimal feeding system for S. cerevisiae, avoiding the production of ethanol to the best of our efforts in order to preserve a high yield (Belo; Pinheiro; Mota et al., 2003). This sought-after model was found in the study by Borowiak et al (2012). The authors had the same goals as us, and described a logistic method for feeding a yeast’s culture in order to maximize biomass production and growth rate. With that in mind, our plan of action was to see if this incredibly well documented study could be an adjustable model to our yeast strain., so we followed their meticulous methodology.

We had high hopes; after all, the authors had previously demonstrated that the logistic model fits very well with the time profile of feeding time when using dissolved oxygen tension (DOT) as a control parameter (Miskiewicz and Borowiak, 2005), and it seemed reasonable to admit that, if we maintained the conditions of our experiment similarly to the authors’, we could test the consistency of this approximation. In doing that, we obtained initial parameters that were applied to the logistic equation and then adjusted using the criterion K. With those results, we were able to calculate the medium’s supply time and, finally, the nutrient medium volume.

So, what results were we looking for? Understanding the balance between nutrient availability and cell growth is essential when determining the optimal time to induce CBD production in yeast without affecting the yield. As seen in our results, we obtained a cell mass curve fairly similar to that of an ideal microbial growth pattern. We can also see a consistent result in our glucose mass, showing a proportional decrease in glucose levels as cell mass increased (as seen in Figure 10).

_{Figure 10: Comparison between consumed glucose and cell mass during the experiments Source:Authors (2024)}

These results and analyses validate our initial hypothesis - that the literature references’ could adjust to our yeast, making it grow in a reasonable profile for what we need. In summary, our results laid a solid and promising foundation for our scale-up studies. This research exemplifies excellence in the application of synthetic biology to produce relevant compounds with the potential to positively impact the industry and the field of biotechnology as a whole, as we use our biological systems to produce a value-added product.

Conclusion

Finally, we know that models and simulations are crucial, especially when it comes to engineering aspects, in order to understand and improve a process as much as possible before applying it. That's why, throughout the year, we focused our energies on developing mathematical models for both 2023 and 2024, in order to consolidate all the theoretical aspects involved in the production of synthetic CBD by yeast. We also made a point of looking for real data, as can be seen in the bioreactor project (See in model), in order to obtain more concrete results linked to a real application of the project and, above all, to develop critical thinking about what can be improved. In addition, we were able to see the progress of the wet-lab during the development of CBDynamics 2024, and this allowed us to compute the bottlenecks and progress of the project, and this was undoubtedly a crucial factor for us to be able to think of viable solutions that propose both innovative optimizations, such as the assembly of RNA interference and overexpression cassettes, but also to understand and analyze what needs to be improved in parts that have already been developed, such as the kinetic modeling stage for the enzymes worked on in 2023.

Human Practices

Excellence should not be understood as synonymous with perfection; rather, it represents the continuous pursuit of achieving the best possible outcomes within established conditions. Despite the application of advanced techniques—such as bioinformatics tools and kinetic modeling—that significantly enhance research and academic endeavors, it is essential to consider the perspectives of affected stakeholders and how these theoretical approaches can be translated into practical solutions. This critical analysis is fundamental for the transition from theoretical knowledge to effective implementation in real-world contexts, thereby achieving excellence in synthetic biology.

In light of this proposal, we sought ways to transcend the confines of academia and engage with diverse communities and unique individuals who could contribute to this solution. We believe that genuine transformation arises from collaboration and dialogue. Consequently, our pursuit of insights and feedback from each sector potentially impacted by our project proved to be crucial. By adopting a proactive approach, we conducted meetings and interviews that not only enriched our understanding but also yielded valuable insights, which guided potential improvements that we strive to implement as effectively as possible.

By actively listening to the voices of impacted parties, we identified opportunities to reflect on our practices and cautiously determine our next steps, ensuring alignment with the objectives of CBDynamics. This approach enables us to promote evolution in responsibility, ethics, and purpose. Through the events organized this year, we successfully raised awareness of our cause, clarifying doubts and addressing prejudices related to the subject matter. This outreach occurred both in academic presentations and through popular media outlets, such as the podcast “MamuCast.”

Following this line of reasoning, we aim to transform the scientific environment through a more comprehensive and inclusive approach. In this context, we integrate synthetic biology across various settings, promoting initiatives in multiple sectors while executing our activities with excellence, supported by relevant research and reflection. To ensure our solutions were genuinely impactful, we developed educational materials and stimulated interest among a significant portion of the population.

Moreover, the convergence of ideas and the quest for inclusion within the academic environment materialized in the creation of inclusive laboratory equipment, facilitating the independence of a growing number of individuals in this domain. This initiative was realized through the development of specialized tools that enhance participation for all. Additionally, the implementation of QR codes for chemical reagents, providing essential information in audio format, contributes to transforming the scientific environment into a truly integrative space.

Therefore, it is possible to assert that there has been a relentless pursuit of improvement across all our segments, thus fulfilling our commitment to excellence in synthetic biology. This field of study is intrinsically related to fundamental principles such as quality, responsibility, ethics, collectivity, and inclusion.

REFERENCES

Amen, T., & Kaganovich, D. (2017). Integrative modules for efficient genome engineering in yeast. Microbial Cell, 4(6), 182–190. https://doi.org/10.15698/mic2017.06.576

Chen, Y., Guo, E., Zhang, J., & Si, T. (2020). Advances in RNAi-Assisted Strain Engineering in Saccharomyces cerevisiae. Frontiers in Bioengineering and Biotechnology, 8. https://doi.org/10.3389/fbioe.2020.00731

Chia, S. Z., Lai, Y.-W., Yagoub, D., Lev, S., Hamey, J. J., Nam, C., … Wilkins, M. R. (2018). Knockout of the Hmt1p Arginine Methyltransferase in Saccharomyces cerevisiae Leads to the Dysregulation of Phosphate-associated Genes and Processes. Molecular & Cellular Proteomics, 17(12), 2462–2479. https://doi.org/10.1074/mcp.ra117.000214

Drinnenberg, I. A., Weinberg, D. E., Xie, K. T., Mower, J. P., Wolfe, K. H., Fink, G. R., & Bartel, D. P. (2009). RNAi in Budding Yeast. Science, 326(5952), 544–550. https://doi.org/10.1126/science.1176945

Gagne, S. J., Stout, J. M., Liu, E., Boubakir, Z., Clark, S. M., & Page, J. E. (2012). Identification of olivetolic acid cyclase from Cannabis sativa reveals a unique catalytic route to plant polyketides. Proceedings of the National Academy of Sciences, 109(31), 12811-12816. https://doi.org/10.1073/pnas.1200330109

Hannon, G. J. (2002). RNA interference. Nature, 418(6894), 244–251. https://doi.org/10.1038/418244a

Leow, J. W. H., & Chan, E. C. Y. (2019). Atypical Michaelis-Menten kinetics in cytochrome P450 enzymes: A focus on substrate inhibition. Biochemical pharmacology, 169, 113615. https://doi.org/10.1016/j.bcp.2019.08.017

Llamosi, A., Gonzalez-Vargas, A. M., Versari, C., Cinquemani, E., Ferrari-Trecate, G., Hersen, P., & Batt, G. (2016). What population reveals about individual cell identity: single-cell parameter estimation of models of gene expression in yeast. PLoS computational biology, 12(2), e1004706.

Luo, X., Reiter, M. A., d’Espaux, L., Wong, J., Denby, C. M., Lechner, A., … Petzold, C. J. (2019). Complete biosynthesis of cannabinoids and their unnatural analogues in yeast. Nature, 567(7746), 123–126. https://doi.org/10.1038/s41586-019-0978-9

Mansouri, M., & Berger, P. (2014). Strategies for multigene expression in eukaryotic cells. Plasmid, 75, 12–17. https://doi.org/10.1016/j.plasmid.2014.07.001

Miyamoto, T., Razavi, S., DeRose, R., & Inoue, T. (2013). Synthesizing biomolecule-based Boolean logic gates. ACS synthetic biology, 2(2), 72-82.

Pratt, A. J., & MacRae, I. J. (2009). The RNA-induced Silencing Complex: A Versatile Gene-silencing Machine. Journal of Biological Chemistry, 284(27), 17897–17901. https://doi.org/10.1074/jbc.r900012200

Shiwei Wang, Juan Xie, Jianfeng Pei, Luhua Lai, CavityPlus 2022 Update: An Integrated Platform for Comprehensive Protein Cavity Detection and Property Analyses with User-friendly Tools and Cavity Databases, Journal of Molecular Biology, Volume 435, Issue 14, 2023, 168141, ISSN 0022-2836,https://doi.org/10.1016/j.jmb.2023.168141

Stout, J. M., Boubakir, Z., Ambrose, S. J., Purves, R. W., & Page, J. E. (2012). The hexanoyl-CoA precursor for cannabinoid biosynthesis is formed by an acyl-activating enzyme in Cannabis sativa trichomes. The Plant Journal, no-no. https://doi.org/10.1111/j.1365-313x.2012.04949.x

Ursula, Oliveira, F., & Giugliani, R. (2024). Atualizando o código genético : o RNA de interferência - um comentário sobre o Prêmio Nobel de Medicina e Fisiologia de 2006. Ufrgs.br. https://doi.org/0101-5575

Wiles, D., Shanbhag, B. K., O’Brien, M., Doblin, M. S., Bacic, A., & Beddoe, T. (2022). Heterologous production of Cannabis sativa-derived specialised metabolites of medicinal significance – Insights into engineering strategies. Phytochemistry, 203, 113380. https://doi.org/10.1016/j.phytochem.2022.113380

Yan Y, Tao H, He J, Huang S-Y.* The HDOCK server for integrated protein-protein docking. Nature Protocols, 2020; doi: https://doi.org/10.1038/s41596-020-0312-x

Yang, X., Matsui, T., Kodama, T., Mori, T., Zhou, X., Taura, F., ... & Morita, H. (2016). Structural basis for olivetolic acid formation by a polyketide cyclase from Cannabis sativa. The FEBS journal, 283(6), 1088-1106. https://doi.org/10.1111/febs.13654

Zirpel, B., Degenhardt, F., Martin, C., Kayser, O., & Stehle, F. (2017). Engineering yeasts as platform organisms for cannabinoid biosynthesis. Journal of Biotechnology, 259, 204–212. https://doi.org/10.1016/j.jbiotec.2017.07.008