Engineering Success

An illustration of the DBTL Cycle:

The Engineering Design below, represents the ideal that our team strives toward, the first step of the DBTL Cycle. However, the point of the cycle is that things become modified in contact with reality. (To see how the Engineering turned out this cycle check out the Results)

During the project many successful project milestones were completed and built, but may have struggled when put to the test, others behaved much better than could have been expected! We’ve built, tested, and learned our way through this project, but let’s look at a specific example and see what we’ve learned.

In theory It is possible to design a process that is perfectly planned, but not with such a complex project and such a small lab team. With only 3 students approved by the university to work in the lab, and only 2 available for much of the summer we had to get creative with our the implementation of or engineering and design.

Finishing the long road to completed golden gate cloning of our multi-plasmid system by the end of August. Knowing our limitations we originally tried to avoid cloning entirely by working with a corporate sponsor for complete plasmid synthesis, but they couldn’t accommodate us. Speaking to our PI’s the lab team calculated 2-5 days for transformation, then 2-5 days for competency, per plasmid. With 4 plasmids total per strain, we would have to remove a plasmid with Lox, all while fighting drawdown effects from the Pichia. In total an additional 2-3 weeks, if everything was successful. A potentially feasible amount of time, our professors advised. Sometimes the willingness to take a shot in the dark pays off however. While performing the sequential experiments as planned, our wet lab captain Simon decided to see what would happen if you just transformed a strain with all 3 plasmids at once. We knew that uptake of DNA was incredibly rare, so it happening 3 times to the same yeast seemed almost impossible, but it only had to happen once. 1 week later we had seeming success! A colony growing on our multi-antibiotic plate! But how to verify transformation?

According to our literature research at the time, the knockout of the gene that catalyzes the final step of the ergosterol pathway (ERG5) was a prerequisite for steroid production. To that end, we designed a process for knocking out this gene through colony PCR. The plan was to design PCR primers to extract flanking regions and then fuse them with a selection marker cassette. In the end we were not successful in performing the ERG5 knockout in the time we had because of issues with the fusion cassete. That doesn’t mean we didn’t learn from the experience. After a seemingly successful multi-plasmid transformation step we needed a way to verify transformation. Remembering our experiences with primer design, we knew colony PCR could be the answer! Designing primers was easy with our previous experience, but again we were frustrated by circumstance. Previous primers had arrived within a week, but while performing our final order with IDT we added extra enzymes which have delayed the process enough that as of wiki freeze they have not arrived. In the end I guess we can learn from that too!


Souza and colleagues1 have replaced ergosterol, which has essential function in modulating membrane physical properties, membrane protein stabilisation and localisation, as well as physiological functions such as endocytosis, protein sorting, and support receptor protein activities in P. pastoris, with cholesterol, a biological and biochemical functional counterpart of ergosterol in animalia 1,2. In doing so, they deleted Erg6 (sterol 24-C-methyltransferase), Erg5 (sterol 22-desaturase) genes, and inserted DHCR7 (Danio rerio 7-dehydrocholesterol reductase) and DHCR24 (Danio rerio delta24-sterol reductase) genes. Importantly, they have shown that their erg5Δerg6ΔDHCR7+DHCR24+ strains are stable and able to produce cholesterol, with a yield of about 1 mg cholesterol per gram wet weight of P. pastoris, using alkaline methanolysis, petroleum ether extraction, and C18 HPLC purification. Could we go further and make it through the steroid pathway up to testosterone? Another important work was the iGEM team 2018 UCSC, who synthesized progesterone de novo aiming at making birth control more accessible. An interesting feature of their work was a riboswitch that would limit the end concentration of the hormone in the fermentation broth and thus mitigate the risk of overdosage upon misuse. An important foundation for their progesterone synthesis pathway was the research done by Duport et al. 1998, who described de novo progesterone synthesis in S.cerevisiae. Since progesterone is a precursor of testosterone, we could build on this already explored synthesis pathway, leaving only the further path from progesterone to testosterone uncharted waters, or at least we thought so initially. Later we learned that Szczebara et al. (2003) have reported the first total de novo synthesis of the steroid hydrocortisone from simple carbon sources in S.cerevisiae, building on the previously by Duport et al.(1998) reported de novo progesterone synthesis process. Their hydrocortisone process would star two of the three enzymes required for testosterone synthesis, leaving only the branch to testosterone open – an enzymatic conversion that we later learned that has already been reported in a stand-alone application as well (Shao et al., 2016). Our work would therefore recombine the work of those publications into a novel path. What further distinguishes our approach from previous work is as follows: While the iGEM group 2018 UCSC aimed at low concentrations of progesterone, we, the iGEM BOKU team 2023, aim at optimizing steroid concentration to make a competitive industrial process for testosterone. Therefore, we not only inserted the enzymes required to produce the hormone, but also aimed to explore the strain development of Duport et al. 1998 that involved the deletion of a native yeast enzyme, thus not having the yeast’s native ergosterol but also its Δ22 saturated form as starting point for sterol synthesis. As a third strain development strategy we planned to use the cholesterol process reported by Souza et al. in 2011 and by Hirz et al. 2013 as basis for the further steroid pathway – a strain development option that has not been reported so far to the best of our knowledge.


Host organism

As host organism, we chose the yeast Komagataella phaffii (formerly Pichia pastoris), strain number CBS7435, kindly provided by BOKU. Up to date, a commonly used organism for production of steroid hormones has been Mycobacterium sp . Mycobacteria can take up sterols such as cholesterol from the environment and grow on them as carbon source, as well as have been shown to natively convert cholesterol and other sterols into steroids. Fernández-Cabezón et al. have used Mycobacterium smegmatis for testosterone production starting from phytosterols, from cholesterol, and from androst‐4‐ene‐3,17‐dione (AD). With M. smegmatis they had chosen a Mycobacterium strain without native ability to metabolize or degrade the AD and testosterone, so that the products would be stable, and inserted the gene for HSD17B, which converts AD to testosterone, recombinantly . The strain of our choice, K.phaffii likewise does not possess the set of enzymes to metabolize AD and testosterone according to the KEGG pathway , except for a hypothetical protein that might convert AD to androstanedione. While mycobacteria are used for their ability to take up cholesterol from the environment, we aimed to avoid the need for cholesterol feeding by exploiting the yeast’s endogenous sterol production. This allows for using simple, cheap and abundantly available carbon sources such as glucose and beyond. This would not be possible in bacteria such as Mycobacterium sp. due to the lack of endogenous sterols in bacterial membranes in general (Huang and London 2017) . To decide which yeast strain we would take, first, we considered S.cerevisiae because it was used by Souza et al (2011) , who succeeded in giving the organism a cholesterol containing membrane. Another consideration was Y. lipolytica, inspired by the iGEM team 2018 of UCSC . Eventually we chose K.phaffii to be our workhorse, since the workgroup of our iGEM supervisors has a strong track of using K.phaffii and would be able to provide us with a variety of resources such as K.phaffii strain CBS7435, Golden Gate cloning plasmids with genome integration sides for K.phaffii, respective primers, and more. Another factor endorsing the choice of K.phaffii was that upon further literature research, we came across the paper of Hirz et al. (2013) who showed that the membrane of K.phaffii can be modified towards replacement of ergosterol by cholesterol in the yeast’s membrane as well, similar to the work of Souza et al. 2011. This was information for us because it shows that our strains would be able to survive with modified sterol content in the membrane. Further, Shao et al. 2016 have described the transformation step of AD to testosterone by insertion of a HSD17B gene into K.phaffii (formerly P.pastoris), proving that the yeast would grow in presence of testosterone. Therefore, K.phaffii would be a good option for the scope of our project.

Experimental design

Our process comprises nine heterologously expressed enzymes organized on three inserted plasmids, further referred to as Module 1, Module 2, and Module 3, as well as two gene deletions (Fig.1a). A detailed overview about the strain development can be seen in Fig.1b,c, while further elaborations of the choice of genes and gene deletions is stated below.

Module 1 genes

In our process, in addition to an established stable cholesterol producing platform (erg5Δerg6ΔDHCR7+DHCR24+) by Souza et al.1, other groups (see Fig.1b, c, f and below), were evaluated. The establishment comprises 3 alternatives in total: 1a), erg5Δerg6ΔDHCR7+DHCR24+, as described and characterised in Souza’s 2011 paper1; 1b), erg5Δerg6+DHCR7+, as described and characterised in Duport et al.3 in Saccharomyces cerevisiae; and 1c), erg5+erg6+DHCR7+ as attempted by the University of California Santa Cruz 2018 iGEM team (see Table.1) in Yarrowia lipolytica. To make those strains, two different versions of the Module 1 genes were constructed: Module 1a comprising DHCR7 and Module 1ab comprising both DCHR7 and DHCR24. cDNA was taken from danio rerio because this has been shown by Souza et al 2011 to work best out of several options in S.cerevisiae, and was subsequently also shown to work in K. phaffii by Hirz et al 2013 .

Zymosterol is the shared precursor of ergosterol (Erg6, Erg2, Erg3, Erg5, Erg4, referred to as Erg2-6 pathway below) and cholesterol (EBP, SC5DL, DHCR7, and DHCR24 pathway Fig.1f)2, consistently, Gu et al.4 advised that ergosterol synthesis pathway competes with exogenous cholesterol biosynthetic pathway. Importantly, erg5Δ and erg6Δ strains are viable4,5,6,7. Thus, our strain A would have enhanced cholesterol biosynthesis given that Erg2 (Δ8, Δ7 isomerisation) and Erg3 (C5 desaturation) share the same function with EBP and SC5DL, respectively. The sole deletion of ERG5 in Erg2-6 pathway was argued to enhance pregnenolone production by minimising desaturation at the carbon 22, making a characteristic Ergosta-5-ene-ol, which is thought to be a better substrate for the module 2 p450scc (CYP11A1)3,8. Thus, in our strain B, Ergosta-5-ene-ol instead of cholesterol serves as a substrate of CYP11A1 (module 2) for pregnenolone synthesis (Fig.1f). Duport and colleagues reported that pregnenolone yielded poorly without erg5 deletion, because Erg5 would convert a part of Ergosta-5-ene-ol to Ergosta-5,22-diene-ol, which they reasoned might not be accepted as substrate for CYP11A13. Thus, our strain C alternative might reasonably show lower pregnenolone, therefore testosterone yield.

Module 2 genes

The products of our first module alternative gene groups (either ergosta-5-ene-ol, or cholesterol) serve as substrate for the second gene module, which comprises CYP11A1 (p450), adrenodoxin, and adrenodoxin reductase (Fig.1f). CYP11A1, also called p450, is a cytochrome monooxygenase that sequentially hydroxylate the C22 and C20 of the cholesterol side chain and eventually cleaves it, leaving behind a keto group at C209,10, known as the side chain cleavage (SCC) activity. The resulting pregnenolone is the key precursor for all steroid hormones in vertebrates. While adrenodoxin serves as a cofactor for CYP11A1, adrenodoxin reductase regenerates reduced adrenodoxin, which is required in the pregnenolone synthesis, say, forward direction9. Ergosta-5-enol, a characteristic intermediate resulting from erg5 deletion (desaturation at C22) and DHCR7 heterologous expression (C7 desaturation), differs from cholesterol, a natural CYP11A1 substrate, in the methyl-group at the C24 position. Duport and colleagues have clearly shown that the side chain of nonconventional ergosta-5-enol is cleaved by CYP11A13. This is further supported by Mast et al., who showed that the side chain binding pocket seem to have allowed structural variations10. However, comparing pregnenolone and peripheral steroid intermediate metabolites yield (as control) between strain A and B may reveal kinetic insights. To construct module 2, cDNA has been taken from bovine origin since Duport et al 1998 have also used cDNA from bovine origin.

Module 3 genes

HSD3B1, CYP17A1, and HSD17B3 were stated on the KEGG webpage’s “steroid hormone biosynthesis” chart to catalyse the transformation from pregnenolone to testosterone. Therefore, the corresponding human cDNA was used for module 3. The choice to take human cDNA was made in lack of reference publications at the time we had to start the experimental part and order the cDNA, so as a first hit we tried human enzymes since we were aiming to construct a molecule that occurs in humans. Later we saw that Szczebara et al 2003 have successfully used the bovine version of CYP17A1 and HSD3B1 to produce hydrocortisone in S.cerevisae, while Shao et al. had been successful in expressing human HSD17B in P.pastoris and using it for efficient biotransformation of 4-androstene-3,17-dione (AD) to testosterone. The question of which organism to take the enzyme from is an interesting one; yet to optimize our industrial process with regards to choice of enzyme homologues, a follow-up project would be needed.

Experimental procedure

Construction of knockout strains

Having a wild type Komagataella phaffii CBS7435 strain and an erg6 knockout strain provided by BOKU, we planned to delete the erg5 gene using PCR to construct a split marker deletion cassette consisting of an antibiotic resistance gene flanked with genomic sequences. Since the erg6 knockout strain already came with an antibiotic resistance, we first had to remove the antibiotic resistance out of the erg6 knockout strain, since we otherwise would have needed more different antibiotic resistance genes than we had at hand. Luckily, we learned that the antibiotic resistance in the erg6 strain had loxP sites, and we succeeded in using Cre recombinase to remove the antibiotic resistance gene. However, we did not succeed in deleting the erg5 gene: Construction of the deletion cassette failed at the stage of fusion PCR that should combine the resistance gene with the genomic flanking sides, and even a thorough analysis did not reveal the reason, since other PCR experiments worked perfectly. As a next step the PCR protocol could have been reviewed, but time did not allow for that. Therefore, we proceeded with the wild type strain and the erg6 knockout as basis for our strain development.

Plasmid construction

The plasmid modules were constructed as follows: Using Golden Gate cloning, we recombined each gene with a housekeeping P. pastoris promoter (either pGAP, pTEF or pMDH3) and a terminator (either rp12aTT, rpp1bTT, or rps25aTT) and recombined all genes of the same module into a module plasmid (BB3) which contains the genomic homologous recombination sites for P. pastoris. These BB3 plasmids were firstly transformed into E. coli cells for amplification before been transformed into Komagataella phaffii (formerly Pichia pastoris) strains: Module 1a (DHCR7) was inserted in the wild type strain and Module 1b (DHC7 + DHCR24) was inserted in the erg6 knockout strain, while Module 2 and Module 3 were inserted in both of the strains as planned.

Analytical methods

Quality and quantity of our intermediates and final product were evaluated as follows: The presence of transformed genes was validated using colony PCR. Finally, the successful production of testosterone and its intermediates was validated via HPLC with a known standard. Quantitation of intermediates such as cholesterol and pregnenolone will allow identification of rate limiting steps and hence narrow down possible targets for future trouble shoot or refinements (such as dHeletion, duplication, down or up-regulation of the target). In the future, the product identity could further be verified by NMR and western blot could confirm the presence of the inserted enzymes.

FIG.1. a), Experimental strategy of our bioengineering process comprises strain development, integration (transformation) of module 1, 2, and 3 genes, and analytical procedure for characterisation, quantitation and identification for future improvement targets. b)-e): Goal plasmids obtained vis Goldengate serial cloning b) and c): module 1 alternatives (see text for details); d), module 2 genes and e), module 3 gene. f), sterol biochemical synthetic pathway relevant to our process. The end product of module 1 gene (mixture of endogenous [purple] and heterologous [blue] genes) are either ergosterol (endogenous or side product), ergosta-5-monoenol (Duport’s process)3, and cholesterol (Souza’s process)1. The product of second module should be pregnenolone as the different side chains of ergosta-5-monoenol and cholesterol are both cleaved. The ergosterol will also be processed into an epoxide side product due to unsaturated (C24) side chain11.


1. Souza, C. M. et al. A stable yeast strain efficiently producing cholesterol instead of ergosterol is functional for tryptophan uptake, but not weak organic acid resistance. Metab Eng 13, 555–569 (2011).
2. Desmond, E. & Gribaldo, S. Phylogenomics of Sterol Synthesis: Insights into the Origin, Evolution, and Diversity of a Key Eukaryotic Feature. Genome Biol Evol 1, 364–381 (2009).
3. Duport, C., Spagnoli, R., Degryse, E. & Pompon, D. Self-sufficient biosynthesis of pregnenolone and progesterone in engineered yeast. Nat Biotechnol 16, 186–189 (1998).
4. Gu, Y., Jiao, X., Ye, L. & Yu, H. Metabolic engineering strategies for de novo biosynthesis of sterols and steroids in yeast. Bioresour Bioprocess 8, 110 (2021).
5. Skaggs, B. A. et al. Cloning and characterization of the Saccharomyces cerevisiae C-22 sterol desaturase gene, encoding a second cytochrome P-450 involved in ergosterol biosynthesis. Gene 169, 105–109 (1996).
6. Palermo, L. M., Leak, F. W., Tove, S. & Parks, L. W. Assessment of the essentiality of ERG genes late in ergosterol biosynthesis in Saccharomyces cerevisiae. Curr Genet 32, 93–99 (1997).
7. Abe, F. & Hiraki, T. Mechanistic role of ergosterol in membrane rigidity and cycloheximide resistance in Saccharomyces cerevisiae. Biochimica et Biophysica Acta (BBA) - Biomembranes 1788, 743–752 (2009).
8. Lecain, E., Chenivesse, X., Spagnoli, R. & Pompon, D. Cloning by Metabolic Interference in Yeast and Enzymatic Characterization of Arabidopsis thaliana Sterol Δ7-Reductase (∗). Journal of Biological Chemistry 271, 10866–10873 (1996).
9. Simpson, E. R. & Boyd, G. S. The Cholesterol Side-Chain Cleavage System of Bovine Adrenal Cortex. Eur J Biochem 2, 275–285 (1967).
10. Mast, N. et al. Structural Basis for Three-step Sequential Catalysis by the Cholesterol Side Chain Cleavage Enzyme CYP11A1 * . Journal of Biological Chemistry 286, 5607–5613 (2011).
11. Slominski, A. T. et al. Novel activities of CYP11A1 and their potential physiological significance. J Steroid Biochem Mol Biol 151, 25–37 (2015).