Bioengineering of the Marine Diatom Phaeodactylum tricornutum with Cannabis Genes Enables the Production of the Cannabinoid Precursor, Olivetolic Acid

Abstract

1. Introduction

Over a thousand of the molecules produced in Cannabis sativa plants have been reported in the literature, including more than a hundred cannabinoids (CBs) [1]. The antinociceptive and appetite-stimulating properties of CBs have been studied thoroughly, as well as their association with relieving the symptoms associated with different diseases such as multiple sclerosis and the relief of chemotherapy side effects [2,3,4,5]. The most studied CBs are ∆9-tetrahydrocannabinol (THC), which has psychoactive potential, and cannabidiol (CBD), which acts on human endocannabinoid receptors to modulate pain and is already contained in many pharmaceutical products available on the market, such as Sativex and Bediol [6,7,8]. Other therapeutic properties of specific CBs that could be administrated in a purified form are still under investigation and offer a promising alternative for different human pathology applications [5]. In planta, phytocannabinoids are produced in glandular trichomes via fatty acid and terpenoid precursors [9]. The genes and enzymes implicated in the cannabinoid pathway were characterized by several plant biologists shortly after the sequencing of the C. sativa genome [9,10,11,12]. The first step in the CB biosynthetic pathway is the formation of olivetolic acid (OA) from malonyl-CoA and hexanoyl-CoA, through a two-step reaction performed by a type-III polyketide synthase named tetraketide synthase (CsTKS) followed by a cyclase named olivetolic acid cyclase (CsOAC), both of which are localized in the cytosol or cells from the glandular trichomes (Figure 1). The enzymes belonging to the type-III polyketide synthase (PKS) superfamily are responsible for producing a wide array of central frameworks that are found in specialized plant metabolites of significant medicinal value. These metabolites encompass various categories, such as flavonoids, stilbenes, chromones, pyrones, phloroglucinols, resorcinols, xanthones, acridones, and quinolones [13,14]. Examples of PKS include chalcone synthase, which accepts p-coumaroyl-CoA as the starter substrate to catalyze three successive condensations with malonyl-CoA to generate naringenin chalcone [15,16] and stilbene synthase, which generates resveratrol [17]. The polyketide synthase from C. sativa (also named tetraketide synthase or olivetol synthase) was characterized and has been shown to be dependent on OAC for the formation of OA (Figure 1) [9]. The 3D structure of CsOAC is most accurately described as a dimeric α + β barrel (DABB) protein, with similarities to other plant DABBs, especially those implicated in stress responses and to the DABB proteins found in Streptomyces bacteria [9]. The aromatic precursor OA is a resorcinol carrying an alkyl chain (named alkylresorcinolic acid), which is condensed with a monoterpenoid moiety (geranyl diphosphate, GPP) to yield the general structure of the most well-known CBs, as reported in feeding studies of 13C-labeled glucose in the early 1960s [18,19]. Although not all CBs are derived from the OA precursor, it remains the most widely characterized pathway in C. sativa [20]. The prenyl moiety of GPP is transferred to OA by an aromatic prenyltransferase (CsAPT) to form cannabigerolic acid (CBGA) (Figure 1). This core intermediate is the backbone of the remaining reaction, which diverges to form cannabidiolic acid (CBDA) and tetrahydrocannabinolic acid (THCA) through CBDA-synthase (CBDAS) and THCA-synthase (THCAS), respectively (Figure 1), as well as other CBs such as cannabichromenic acid (CBCA) via cannabichromenic acid cyclase (CBCAS) [9,11,21,22]. The neutral phytocannabinoids THC and CBD are generated from cannabinolic acids by non-enzymatic decarboxylation and are often produced upon exposure to light or heat (Figure 1) [21]. Aside from CBs, the cannabis plants also produce other specialized metabolites from shared precursors with cannabinoids, such as fatty acids, competing with olivetolic acid synthesis, and terpenoids such as limonene [23], competing for the GPP pool [24,25].

The yield of CB in plants is limited, both in quality and in quantity; the maximum production is due to a mixture of these metabolites accumulating in the flowers (20 mg/g fresh weight) [6,26]. In addition, the lack of scientific literature on cannabis physiology leads to the absence of standardized culturing protocols, resulting in a variation in CB profiles, quality, and quantity. In addition, cannabis pathogens such as the gray mold causal agent (Botrytis cinerea) affect the overall plant yield and metabolite composition [27]. As well as multiple unknown facets of CB production in planta, this suggests that the implementation of a heterologous expression system with fewer components and fewer competing enzymes might be a solution for achieving a higher production yield of single CB components, such as CBD or THC. Therefore, synthetic biology offers an important alternative method to produce CBs for the pharmaceutical field with less risk and fewer crop management challenges.

Several research groups aimed to design de novo or improve pre-existing cannabinoid production in plant tissue cultures, but apart from the transformation rates, in vitro culture viability and yield are still to be optimized [6]. For instance, few groups succeeded in transforming yeast (Saccharomyces cerevisiae) and bacteria (Escherichia coli) to produce the different metabolites of the CB pathway, such as OA or CBGA [28,29]. The production of these metabolites in yeast and bacteria often involves the addition of costly precursors such as acetyl-CoA, hexanoyl-CoA, or olivetolic acid [28,30], or involves supplementing the media with sugars [31], such as galactose, to provide a more affordable alternative with a long series of gene stacking. For instance, the production of OA in Dyctiostelium discoideum on a 300-L scale led to 4.8 µg/L [32]. OA production in E. coli reached 80 mg/L after introducing a series of modifications to increase the CoAs pool [33]. However, the bacterial and yeast translation and post-translational mechanisms, as well as their metabolic chassis, are distinct from the mechanisms found in higher plants. This resulted in low yields, such as picomoles per unit of optical density, from these microorganisms when compared to the bioengineering and supplementation costs [31,34]. By contrast, diatoms are photosynthetic organisms that share the main metabolic paths with plants, making them promising candidates for the production of heterologous compounds [9]. All the precursor pathways, such as acetyl-CoA, hexanoyl-CoA, GPP, etc., are present in diatom cells at different times of the cell cycle, which reduces the need to supplement the culture media with costly precursors or sugars [35]. Also, the extrachromosomal transformation of diatoms via bacterial conjugation allows the fairly stable expression of heterologous genes [36,37]. Similar advances in the bioengineering of diatoms have been shown recently, allowing the production of monoterpenoids such as geraniol, an insect repellant, in the model diatom Phaeodactylum tricornutum [36].

Thus, the aim of this study is to produce the CB precursor OA in the marine microalga, P. tricornutum. Here, we inserted cannabis CsTKS and CsOAC genes into a stable optimized episome that was assembled in S. cerevisiae [38,39]. The recombinant episome was then transformed via a trans-kingdom conjugation between E. coli and P. tricornutum [38]. The transconjugant strains were screened for their ability to produce molecules of interest, using a high-performance liquid chromatograph coupled to an ultra-violet detector (HPLC-UV). We detected OA (0.6–2.6 mg/L) in P. tricornutum transconjugants that issued from two different expression cassettes. Although the production of the OA compounds of interest was temporary, this work sheds light on a powerful and suitable system for plant metabolite heterologous production in a cost-effective phototropic system employing P. tricornutum.

2. Results

3. Discussion

The bioengineering of P. tricornutum with cannabis genes represents a promising approach to addressing the challenges associated with traditional cannabinoid production. By harnessing the unique capabilities of marine diatoms, this study paves the way for sustainable and controlled cannabinoid biosynthesis. Furthermore, the successful expression of cannabis genes in a non-native host highlights the versatility of synthetic biology in the context of metabolic engineering. This work represents the first report of cannabinoid engineering in brown microalgae and provides a proof-of-concept evaluation of P. tricornutum for the heterologous production of OA and cannabinoids. We provide evidence that the natural metabolism of the marine diatom P. tricornutum is well-suited for use in metabolic engineering to produce the cannabinoid precursor, OA, and CB-like metabolites.

The episomal vectors, PtOA1 and PtOA2, contained both CsTKS and CsOAC genes, separated by a self-cleaving peptide sequence and under the control of the same promoter, to ensure similar levels of expression of both genes and, subsequently, comparable levels of accumulated enzymes for performing the two-step reactions. The transconjugants PtOA1 and PtOA2 only differed in terms of the addition of a 6xHis tag on CsTKS for easier detection of the protein. In PtOA3, both genes were linked together with a linker and tagged with YFP for visualization and localization studies.

Although a previous study demonstrated the occurrence of episomal rearrangements [42], in this study, the P. tricornutum transconjugant strains obtained for each episomal vector did not show significant DNA sequence rearrangement after 1 year (Appendix A, Figure A1 and Table A4). This suggests that episomal rearrangements could be sequence- or DNA-motif-specific. Also, the PtOA1-3 transconjugant strains grew similarly to the WT and EV strains (Figure 3 and Figure 4). This suggests that the presence and expression of the transgenes and the produced metabolites did not significantly affect the growth and division of P. tricornutum. Genome analysis of green algae, diatoms, and higher plants revealed similarities in lipid metabolism [43,44]. Therefore, the fact that there is little to no effect of CsTKS and CsOAC expression on P. tricornutum growth and fitness could be a result of the nature of the pathway, as an extension of fatty acid metabolism that is fairly close to the diatom chassis, and a preference for energy conversion into lipids.

Heterologous protein detection was not consistent among the different transconjugant strains (Figure 5 and Appendix A, Figure A2). For example, the four analyzed transconjugant strains of PtOA1 show a similar pattern of accumulation, whereas, in the PtOA2 strains, the T2A sequence did not appear to be cleaved. This could be explained either by the possible presence of negatively acting cis elements in the sequence of PtOA2 transgenes [43] or by the nature of the amino acid environment that surrounds the T2A sequence cleavage site [45]. It was also reported that the gene following the T2A sequence would be affected on the expression level [46] and, thus, would yield fewer CsOAC transcripts, but the effect of this phenomenon would be validated if the C10–C12 polyketide product was available as a standard for HPLC detection, which is not the case.

When compared to previous studies, the production of OA in Yarrowia lypolytica slightly changed when CsTKS and CsOAC were expressed under different promoters, compared to when the two genes were fused [42]; in P. tricornutum, the PtOA1 showed a higher quantity of OA (Figure A3) compared to there being none in PtOA3 when the two genes were fused by a glycine-serine linker. This is not applicable in PtOA2 since the Western blot did not show cleaved proteins, which could indicate that the cleaved proteins were below detection levels.

To our knowledge, this is the first time that a brown microalgae host has demonstrated the capacity to accumulate OA and use it for the synthesis of CB-like metabolites. The heterologous production of OA was observed in the different transconjugants of P. tricornutum, transformed with episomal constructions and harboring various combinations of C. sativa TKS and OAC genes (Figure 2; Appendix A, Figure A1 and Table A2). The cassette designs in this study were fairly simple and direct, with only two main cannabis genes being introduced. When compared to the literature, the first approaches regarding yeast achieved the production of only 0.48 mg/L of OA by expressing CsTKS and CsOAC, and by feeding the culture with sodium hexanoate [9]. Here, we obtained higher yields (0.56–2.6 mg/kg) compared to yeast with and without hexanoate supplementation [9]. In Y. lipolytica, it was bioengineered with Pseudomonas sp. LvaE, encoding a short-chain acyl-CoA synthetase, acetyl-CoA carboxylase, pyruvate dehydrogenase bypass, NADPH-generating malic enzyme, as well as the activation of the peroxisomal β-oxidation pathway and the ATP export pathway to redirect the carbon flux toward OA synthesis [47]; the yield was also less than 0.5 mg/L until batch culture optimization.

Further optimization of the P. tricornutum transconjugant strains by boosting the pool of precursors could increase the production of OA. In E. coli, besides introducing CsTKS and CsOAC, metabolic engineering consisted of the overexpression of the acetyl-CoA carboxylase subunits A, B, C, and D from the same host to ensure the high production of malonyl-CoA, and a reversal of beta-oxidation that allowed the accumulation of hexanoyl-CoA [33]. This led to a higher quantity of OA (46.3 mg/L) after the optimization of culture conditions and induction parameters [33]. In P. tricornutum, hexanoyl-CoA and malonyl-CoA occur naturally, due to the long-chain acyl-CoA synthase (PtACS3 and PtACS4) for hexanoyl-CoA and to the acetyl-CoA ATP-dependent carboxylase (PtACCase) for malonyl-CoA [43]. P. tricornutum was also shown to accumulate intracellular amounts of GPP, along with neryl diphosphate (NPP) [36,48]; both precursors are donors of the prenyl moiety of several cannabinoids, such as CBGA and cannabinerolic acid (CBNA) in cannabis [46]. However, overexpression of these upstream genes could eliminate the need for supplementation while maintaining the necessary pools of precursors for more stable production of cannabinoids, which is similar to the process used with yeast [31,34].

Along with precursor bioavailability, endogenous enzymes are required to enable the conversion of OA into CB or CB-like compounds, as observed in this study, either by direct production (Figure A4) or by supplementation assays (Table 1). Bioinformatic analyses led to the identification of two putative enzymes (Appendix B, Figure A6) that are encoded by Phatr3_J37858 and Phatr3_J2738.t1 and are capable of prenyltransferase activity, thereby allowing the production of CBGA. Both enzymes are UbiA prenyltransferase family members that could be used originally by the diatom to produce terpenes or flavonoids [49,50]. The in silico identification of THCAS and CBDAS-like endogenous candidate genes is more intricate, given the specificity of these enzymes. Our approach can also be used to further characterize endogenous diatom enzymes with multiple functions by testing the reactivity of diatoms to drug precursors in higher-level plants.

The episomal sequences were well-conserved in the timeframe of this study in all three constructions of interest (in DNA and proteins). However, after such a timeframe, the production of OA was not detectable, perhaps due to silencing via an epigenetic mechanism [42], mRNA or protein stability [37], and, less probably, episomal rearrangements, as seen in other cases [42]. It would be useful to characterize a larger number of transformed P. tricornutum using a method similar to the process used with yeast [31,34], or to try randomly integrated chromosomal expression (RICE). Interestingly, a loss of production was observed in parallel with a change in cell morphotype from a triradiate to a fusiform population, which could also be associated with changes in the intracellular levels of certain precursors related to general metabolism [51]. A loss of heterologous production of specialized metabolites was previously observed for E. coli, where, after the 31st generation, lycopene production decreased from 3.3% of its initial yield [52] due to segregational plasmid instability. Similarly, the use of cyanobacteria as a heterologous production platform faces many challenges, such as a loss of productivity due to the burden or loss of the metabolic pathway and genetic instability [52,53]. In yeast, low expression-cassette instability in large fermentation batches led to a decrease in recombinant protein production [54]. Taken together, these studies show that a loss of production seems to occur elsewhere in other systems, as a result of different underlying problems. In addition, these further demonstrate that like other systems, diatoms need further optimization of the metabolic pathway-encoding constructs discussed below to ensure a suitable platform for the heterologous production of specialized metabolites. Recently, cloning the Metarhizium anisopliae ARSEF23 highly reducing PKS enzyme (MaOvaA), in addition to a non-reducing PKS (MaOvaB) and a domain PKS (MaOvaC) consisting of an acyl carrier protein and a thioesterase, in the fungus Aspergillus nidulans yielded high titers of OA (80 mg/L) [55]. Thus, it would be interesting to compare the expression and behavior of cannabis enzymes regarding fungal enzymes such as highly reducing and non-reducing PKS, along with thioesterase and acyl carrier enzymes [50]. Another approach to increase OA titers could be the construction of alternative pathways using enzymes found in other OA-producing plants than cannabis, in order to achieve OA and other CB precursors while using fungal enzymes or mutation-enhanced enzymes. It is possible, from an evolutionary point of view, that other sources of transgenes are more suitable for heterologous expression in diatoms.

Moreover, in this study, low OA yield could be explained by the detection of cannabinoid-like compounds in higher titers (Figure A4) similar to those observed in other studies, where flavonoid precursors were consumed to form flavonoids in bioengineered yeast [56]. Further steps of compartmentalization of the products would be an interesting strategy, given the possible cytotoxicity of the cannabinoids in many systems [12,57].

The application of P. tricornutum as a production platform for OA raises intriguing questions about its metabolic compatibility with the broader CB biosynthesis pathway. Further research is warranted to understand how the engineered diatom handles the downstream conversion of OA into various cannabinoids. Additionally, optimizing the growth conditions, such as light intensity, nutrient availability, and cultivation scale, will be crucial for maximizing production yields [51,58,59]. The successful synthesis of olivetolic acid in P. tricornutum opens the door to a range of future possibilities. Further pathway engineering could enable the production of specific compounds, offering a renewable and controllable source of cannabinoids for medical and industrial applications. The engineered diatom strains could be integrated into larger bioprocessing systems, harnessing their natural growth advantages for cost-effective production.

4. Materials and Methods

5. Conclusions

In conclusion, the bioengineering of P. tricornutum with cannabis genes represents a significant advancement in CB production technology. This innovative approach not only provides a sustainable alternative to traditional cannabis cultivation but also highlights the potential of marine microorganisms as versatile bioreactors for valuable natural products. As synthetic biology and metabolic engineering continue to evolve, the intersection of marine biology and cannabis biosynthesis could shape the future of CB production.

Acknowledgments

Abbreviations

Appendix A

Our study showed the construction of three episomes in the pPtGE30 backbone containing C. sativa genes for the production of OA (Appendix A, Figure A1 and Table A1, Table A2 and Table A3). Each episome was inserted into P. tricornutum via bacterial conjugation and the corresponding enzymes were successfully accumulated (Appendix A, Figure A2 and Table A4).

Appendix B. The In Silico Analyses Identified Putative Endogenous P. tricornutum Candidates for CB-like Biosynthesis

To investigate the hypothesis that the endogenous enzyme could yield cannabinoids, characterization of the enzyme candidates was performed using bioinformatic tools. Putative candidate sequences for endogenous P. tricornutum enzymes that were able to convert OA and GPP into CBGA were obtained from the HMMER web server [64], using the CsAPT4 amino acid sequence as a query against the Protist database. We obtained five hits where the protein B7G488 (gene ID Phatr3_J37858) (57% similarity) and the protein B7FT51 (gene ID: Phatr3_J2738.t1) (51% similarity) had been annotated as the predicted protein members of the UbiA prenyltransferase family, to which CsAPT4 belongs [49]. The phylogenetic analysis indicated that these two proteins are clustered closer, according to sequence similarity, to the C. sativa APTs group (Appendix A, Figure A5). The soluble aromatic prenyltransferase from Streptomyces sp., NphB, had already been characterized by Kuzuyama et al. in 2005 [65], in a cluster with that from Aspergillus terreus, a fungal APT. No hits were found by using the same approach to identify endogenous diatom-soluble APT candidates, using Streptomyces sp. NphB as a query. Endogenous P. tricornutum candidate enzymes that were similar to THCA/CBDA synthase were not found in the database. One putative hypothetical candidate (with a 25.50% sequence identity compared to THCAS) from the diatom database was the violaxanthin deepoxidase-like protein (VDL), identified for its putative function in pigment deepoxidation and its requirement of a specific pH, which is similar to THCAS.

Funding Statement

This research was funded by the Natural Sciences and Engineering Research Council of Canada through the Alliance Program (award nos. ALLRP 554429-20 and ALLRP 570476-2021) to I.D.-P. Additional support in the form of scholarships was awarded to F.A. from MITACS-Elevation IT15971, to E.I.F. and A.M.D.-G. from the MITACS-Acceleration program (grant nos. IT12310, IT16463, and IT19432), and to I.D.-P. is also acknowledged. B.J.K. is supported by Natural Sciences and Engineering Research Council of Canada (RGPIN-2018-06172).

Author Contributions

F.A., E.I.F., N.M., F.M.-M. and I.D.-P. contributed to the study’s conception and design. Methodology, material preparation, data collection, and analysis were performed by F.A., E.I.F., M.H., A.M.D.-G., N.M., A.C., S.-E.G. and F.M.-M. Episomal constructions, diatom conjugation, and molecular biology experiments were planned and conducted by F.A., E.I.F., A.M.D.-G., J.L., N.M., F.M.-M., B.J.K. and I.D.-P. HPLC analyses were performed by M.H., A.C. and S.-E.G., whereas all LC-MS analyses were performed by J.-F.L. The first draft of the manuscript was written by F.A., E.I.F., A.M.D.-G. and N.M. F.M.-M. and I.D.-P. edited the previous versions of the manuscript. I.D.-P. supervised the project and secured the funding. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

Data is contained within the article.

Conflicts of Interest

The authors declare that this study received funding from Algae-C. The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article or the decision to submit it for publication.

Footnotes

References