Week 1 HW: Final Project Ideas

Project: Biomanufacturing of trabectedin (ET-743 / Yondelis)

(1) What I want to biomanufacture and why:

I want to biomanufacture trabectedin (ET-743, trade name Yondelis), an anticancer drug already clinically approved for soft-tissue sarcoma and recurrent ovarian cancer. It is not a hypothetical compound; it is a real, currently used medicine that saves lives—the issue is not whether it works, but how it is produced. There are two motivating factors. First, the scarcity is extreme: a full ton of wild tunicates is required to isolate just one gram of trabectedin. Second, the source species, Ecteinascidia turbinata, lives on Caribbean mangrove roots—a Latin American ecosystem—yet the drug is produced and controlled almost entirely in Europe. Its cost makes it inaccessible to most healthcare systems in the region where the organism originates. I want to address this issue of biotechnological sovereignty: producing the compound without relying on the harvesting of tons of wild tunicates.

(2) Which living system I would use:

Here is the twist that makes this project truly interesting: the tunicate does not produce the trabectedin itself—a bacterium living inside it does. The biosynthetic pathway was found to depend on an unculturable bacterial symbiont, Candidatus Endoecteinascidia frumentensis—a gammaproteobacterium with an extremely reduced genome (~631 kb), typical of organisms that have evolved in close dependence on a host. My proposed chassis: instead of attempting to culture this bacterium (which has resisted all attempts at independent cultivation so far), I would transfer its biosynthetic gene cluster—a polyketide-non-ribosomal peptide (NRPS) pathway that assembles three tetrahydroisoquinoline units—into a culturable, well-characterized bacterial chassis, such as E. coli or Pseudomonas fluorescens (the latter is, in fact, already used today to ferment a precursor called cyanosafracin B, which is used in the drug’s current semi-synthesis). The system’s input consists of the metabolic precursors required by the NRPS pathway (amino acids, fatty acids); the process involves the sequential enzymatic assembly of these building blocks; the output is the molecule with its seven chiral centers correctly formed—hence why enantiomeric purity is the critical quality control point for the entire process.

(3) How I would use automation:

I would automate two distinct stages of the workflow:

Prototyping: The C. Endoecteinascidia frumentensis biosynthetic cluster is large (a ~35 kb core, with the complete cluster spread across more than 173 kb of the symbiont’s reduced genome). Therefore, testing which combination of promoters, RBSs, and codon-optimized variants actually expresses in a cultivable chassis (E. coli or P. fluorescens) requires constructing and testing many versions in parallel. I would use a liquid-handling robot (such as an Opentrons) for combinatorial DNA assembly and plate-based micro-cultivation, coupled with an automated optical sensor for circular polarization readings (revisiting the earlier “chiral light” concept) to rapidly check enantiomeric purity in each clone, avoiding reliance on costly chiral HPLC for every sample. Scaling: Once the best strain is identified, fermentation would take place in a bioreactor with automated control of pH, dissolved oxygen, and feed rates. This utilizes the same type of infrastructure currently used in industry to ferment Pseudomonas fluorescens for producing cyanosafracin B—the precursor for the current semi-synthesis method—so this is not process science fiction; it is an extension of a system already operating at full scale.

(4) Quantity and timeframe:

(a) Quantity: I am using Chile as a reference scale. The clinical dose is 1.5 mg/m² of body surface area per cycle, administered every 21 days; given an average adult body surface area of ​​~1.8 m², this amounts to 2.7 mg per cycle. Between 30% and 50% of patients complete 6 or more treatment cycles, so I am taking 6 cycles as a representative treatment course: ~16.2 mg per patient. Chile lacks its own registry for soft-tissue sarcoma incidence—a study by the Universidad Austral de Chile explicitly notes this and uses international rates as a proxy—so I do the same: applying the U.S. rate (~13,400 cases/year in a population of ~333 million, or ≈4 per 100,000/year) to Chile’s population of ~19.5 million, I estimate ~780 new cases of soft-tissue sarcoma annually. This results in an annual requirement of ~12.6 grams of trabectedin to cover one treatment course per patient in Chile. To illustrate the absurdity of the current method: since one ton of wild tunicates is required for every gram produced, meeting only the Chilean demand via wild harvesting would require ~12.6 tons of tunicates per year—an unfeasible amount. (b) Timeframe: Here, I introduce a significant assumption that I will explicitly label as such: I assume a conservative initial production titer of 1 mg/L in the heterologous host (a reasonable figure for a first-generation expression system of such a complex NRPS pathway, which has never before been reconstituted outside its original symbiont). At that titer, Meeting the annual requirement of 12.6 g would necessitate ~12,600 liters of culture—an amount achievable in just a couple of runs using a 5,000–10,000 L industrial bioreactor; in other words, weeks of fermentation spread across a few batches per year, rather than years of cultivation.

(5) What I don’t know yet:

Three genuine technical gaps, not merely formalities: Whether the pathway functions outside its host. The symbiont C. Endoecteinascidia frumentensis has never been cultured independently; it possesses an extremely reduced genome, typical of organisms that rely on their host for functions lost over the course of evolution. I do not know if the tunicate provides a specific cofactor, precursor, or cellular machinery without which the NRPS pathway simply fails to assemble the molecule in a foreign chassis like E. coli. This would need to be determined experimentally by expressing the pathway in modules and observing where it stalls. Whether the actual yield bears any resemblance to my assumed 1 mg/L. I came up with that figure as a conservative starting point—there are no published reports of functional heterologous expression of this complete pathway outside the original symbiont. The actual yield could be 100 times lower (or initially nonexistent), which would completely alter the scale-up calculations in section 4. Whether the circularly polarized light sensor can actually detect enantiomeric purity at the low concentrations typical of early-stage cultures. While the phenomenon of circularly polarized bioluminescence has been demonstrated in fireflies, I do not know if such a sensor is sensitive enough to distinguish between enantiomers at the microgram-per-liter level; this would need to be validated using pure chemical standards before relying on it for quality control in the bioreactor.

References:

1. Rath, C.M., Janto, B., Earl, J., et al. “Meta-omic Characterization of the Marine Invertebrate Microbial Consortium That Produces the Chemotherapeutic Natural Product ET-743.” ACS Chemical Biology, 2011, 6(11), 1244–1256.
2. Schofield, M.M., Jain, S., Porat, D., Dick, G.J., Sherman, D.H. “Identification and Analysis of the Bacterial Endosymbiont Specialized for Production of the Chemotherapeutic Natural Product ET-743.” Environmental Microbiology, 2015.
3. Cuevas, C., Francesch, A. “Development of Yondelis (trabectedin, ET-743). A Semisynthetic Process Solves the Supply Problem.” Natural Product Reports, 2009, 26(3), 322–337.
4. Sherman, D.H., Williams, R.M., Ehrlich, G.D., Rath, C.M., et al. “Biosynthetic Pathway for Heterologous Expression of a Nonribosomal Peptide Synthetase Drug and Analogs.” U.S. Patents 8,815,562 B2 (2014); 9,611,491 B2 (2017); 9,994,885 B2 (2018).

Project developed with the support of Claude (Anthropic) for brainstorming, research, technical data verification, and drafting the initial version.

Final Project