Week 1 HW: Final Project Ideas

Project Idea: Cell-Free Manufacturing of chemotherapeutic Etoposide Precursor

(1) What you would like to biomanufacture and why?

The goal is to biomanufacture podophyllotoxin, a natural precursor used in the production of the chemotherapeutic drug etoposide, which is used primarily to treat testicular cancer and small-cell lung cancer. Podophyllotoxin is traditionally obtained from mayapple plants, particularly Podophyllum hexandrum (Himalayan mayapple), which contains higher concentrations of podophyllotoxin than the more common American mayapple, Podophyllum peltatum. The reliance on these plants for extraction creates both a supply challenge and a conservation concern, particularly for P. hexandrum, which has experienced significant pressure from harvesting. There has already been work toward producing podophyllotoxin in alternative biological systems, including engineered tobacco plants. However, I am interested in exploring whether a more controllable biomanufacturing platform, such as a microbial chassis or cell-free system, could eventually provide a more flexible and scalable route to production. My aim is to both: contribute to a more reliable supply of an important pharmaceutical precursor, potentially making its downstream production more accessible and less dependent on plant harvesting, while also reducing pressure on the native plant populations from which podophyllotoxin is traditionally obtained.

(2) What living system would you use?

E. coli would be used as the production host for the enzymes needed to build the cell-free system. The genes encoding the mayapple pathway enzymes, including pinoresinol-lariciresinol reductase (PLR), secoisolariciresinol dehydrogenase (SDH), two O-methyltransferases (OMT1 and OMT3), the ring-closing 2-oxoglutarate dioxygenase (2-ODD), and the three cytochrome P450s (CYP719A23, CYP71CU1, and CYP82D61), would be synthesized and introduced into separate E. coli strains, with each strain expressing one enzyme or a small group of enzymes.

One thing I would have to account for is that plant P450s can be difficult to express as active enzymes in E. coli because they normally require a membrane environment and an electron-donor partner. However, Decembrino et al. demonstrated that these three specific mayapple P450s can be expressed and function in E. coli when co-expressed with the plant-derived NADPH reductase ATR2 from Arabidopsis. Because of this, I would use that approach rather than switching to a eukaryotic host such as yeast or insect cells.

After expression, the E. coli cells would be lysed and the enzymes collected, either as purified enzymes or as semi-crude lysates. These enzyme fractions could then be combined in a single cell-free reactor with the starting substrate, coniferyl alcohol, as well as the cofactors needed for the different reactions. These would include SAM for the methylation steps, NADPH for the P450 reactions, and Fe²⁺, 2-oxoglutarate, and O₂ for the ring-closing reaction.

The idea is essentially to take the pathway out of the cell. Instead of having the engineered E. coli grow and try to produce the final molecule themselves, I would use them to produce the individual enzymes and then combine those enzymes in a controlled cell-free system. The enzymes would act sequentially on the starting substrate until the pathway reaches the desired product, 4’-demethylepipodophyllotoxin, which could then be purified from the reaction mixture.

(3) How would you use automation?

An Opentrons liquid-handling robot would mainly be used for the optimization and screening stage of the cell-free reactor, before moving to a larger-scale system.

I would use the Opentrons to:

Set up the reactions: dispense different combinations and ratios of the ~8 enzyme fractions, along with different concentrations of the required cofactors, into a 96-well plate. This would allow me to run many small reactions in parallel instead of manually pipetting every condition.
Take samples: automatically collect small samples from each reaction at set time points so I could see how the pathway is progressing.
Measure the results: transfer those samples to a plate reader or HPLC/LC-MS system to measure product formation.
Optimize the conditions: use the results to guide a Design-of-Experiments (DoE) approach, testing different enzyme ratios, cofactor concentrations, pH, temperature, and other conditions. The robot could then set up the next round of experiments based on the results from the previous one. The main reason I would use automation here is that there are a lot of variables to optimize. Once I start combining ~8 enzymes with different enzyme ratios, cofactor concentrations, pH, temperature, and reaction times, the number of possible conditions becomes very large. Having the Opentrons set up these reactions would let me test many more conditions in parallel and make the optimization process much faster and more systematic.

I would use the Opentrons as a prototyping and optimization tool, rather than trying to use it for the final production scale. It is well suited for small-volume, highly parallel experiments, while scaling the optimized reaction to liters or more would require a different system, such as a stirred-tank or flow reactor. I would therefore use the Opentrons to figure out what works, and then transfer those conditions to the larger production system.

(4) How much would you need to make, and how long would it take? (a) How much?

I would set the production target based on enough etoposide to cover the two FDA-approved US indications I am considering: testicular cancer and small-cell lung cancer (SCLC) for one year.

1. To estimate the number of patients

Testicular cancer: 9,720 new US cases/year
SCLC: 13% of 226,650 annual US lung cancer diagnoses ≈ 29,465 cases/year
Total ≈ 39,185 patients/year or 39,200 patients/year

2. To Estimate etoposide required per patient

For a rough estimate, I used standard treatment regimens and an average adult body surface area of 1.9 m².

Testicular cancer: 100 mg/m² × 1.9 m² × 3 doses/cycle × 4 cycles = 2,280 mg ≈ 2.3 g/patient
SCLC: 100 mg/m² × 1.9 m² × 4 doses/cycle × 5 cycles = 3,800 mg ≈ 3.8 g/patient

If we use around 3 g etoposide per patient as a rough average,

39,200 patients × 3 g = 117,600 g

then we will have to produce 117,600 g of the drug or 118 kg etoposide/year total.

3. Convert this to the amount of aglycone that needs to be produced

The cell-free pathway would produce the etoposide aglycone rather than etoposide itself. The final sugar-attachment step would be carried out separately.

If I assume an overall 50% yield after the enzyme cascade, purification, and final chemical glycosylation:

118 kg ÷ 0.50 = 236 kg

THen the estimated production target would be around 236 kg precursor/year.

(b) How long?

For the production time, I would use the results from Decembrino et al. as a reference point. Their work reached 137 mg/L for an early pathway intermediate and reported up to 98% yield for a five-enzyme cascade to deoxypodophyllotoxin. Since the full pathway I am proposing would involve additional steps, I would use more or less 100 mg/L as a deliberately rough starting assumption for the final aglycone.

If we made the following assumptions about production

Titer: ~100 mg/L Reactor volume: 1,000 L Batch time: ~2 days

In a cloed batch reactor that would be, per batch:

100 mg/L × 1,000 L = 100,000 mg → 100 g aglycone/batch

Comparing that with the annual target:

236 kg ÷ 0.100 kg/batch = 2,360 batches

At one reactor running continuously:

2,360 batches × 2 days = 4,720 days

4,720 ÷ 365 ≈ 12.9 years

One 1,000 L reactor would take ~13 years. To produce the same amount in one year we would need around 13 reactors running in parallel.

(5) What don’t you know yet?

There are still a few important things I would need to figure out before knowing whether this system would actually work at the scale I am proposing:

Real titer and yield for the full pathway: The literature numbers I used are from a 5-enzyme partial cascade that reaches deoxypodophyllotoxin, rather than the full ~8-enzyme pathway to the aglycone. I don’t know yet how much the additional steps would reduce the overall yield. For example, if the actual titer were closer to 10 mg/L instead of my assumed 100 mg/L, the number of reactors needed would increase by roughly 10x.
P450 stability and reusability: I am assuming that the enzymes, particularly the P450s, would remain active throughout a roughly 2-day batch. I don’t know if that is realistic in a cell-free system. I would need to test their stability and determine whether they can be used for a full batch, or whether an approach such as enzyme immobilization would be needed to extend their lifetime.
Efficiency of the final glycosylation step: My 50% overall process-yield assumption is mainly a placeholder for estimating the scale of the system. I would need to determine the actual yield of converting the purified aglycone into etoposide and understand how much of this step is already established at industrial scale. This could significantly change the amount of aglycone I would actually need to produce upstream.

Overall, the biggest unknown is whether the full pathway can reach a useful titer while keeping the enzymes stable enough for a practical cell-free process. Those are the first things I would want to test experimentally before treating the reactor and production estimates as realistic.

AI Client: ChatGPT & Claude

Prompts:

What does automation actually mean in a biomanufacturing project? How could I make automation meaningful rather than just having a robot dispense things? Could I use a cell-free manufacturing system instead of a living organism? What would be a good cell-free + Opentrons project? I want something with a strong medicine/cancer angle.
I want something analogous to the Madagascar periwinkle/cancer-drug story but with a simpler pathway. Are there other plants that help other diseases that I could try to make the active compound? What plant-derived drugs have relatively simple biosynthetic pathways?
Are there plant-derived medicines that could be made with ~1–5 enzymes? Which of those would actually work for a cell-free system? Which ones would have a good Opentrons automation angle? What about medicines from plants other than the obvious cancer ones?
Could I make the active compound rather than the whole plant extract? What are some plant → active compound → disease examples?
What about L-DOPA? What about 5-HTP?What about galantamine? What about colchicine? What about artemisinin? What about camptothecin? What about podophyllotoxin? What about etoposide? What about Taxol/paclitaxel?What about vinblastine/vincristine from Madagascar periwinkle? What about cyclopamine?
What about marine-derived compounds instead? Could I use a marine enzyme to make a pharmaceutical compound?
What about a marine brominase/halogenase? What about a marine-derived enzyme that makes a pharmaceutical precursor?
Moving toward mayapple Is mayapple a good one? What exactly does mayapple make? Is podophyllotoxin itself the drug, or is it a precursor? What cancer drug comes from podophyllotoxin? How complicated is the podophyllotoxin pathway? Can I isolate a 3–5 enzyme section instead of doing the whole pathway? Is there a small enzyme module within the mayapple pathway? Could I make deoxypodophyllotoxin instead of the complete drug? Could I use the pathway toward etoposide rather than trying to make etoposide itself?
What did the 2015 mayapple paper actually do? Since that 2015 paper, has the tobacco idea gone anywhere? Did they actually make the compound in tobacco? How much did they make? Did later papers improve the tobacco system? What happened after the 2015 pathway-reconstruction paper? Did they identify bottlenecks? Did they improve precursor supply? Did they eventually make more of the etoposide-related compounds? Could the tobacco system be replaced by a microbiological chassis? Could it be replaced by a cell-free system? Would it be correct to say that a microbial or cell-free system could improve on tobacco? How much of the actual drug is now made using tobacco versus the original plant? Is tobacco actually being used commercially to make etoposide? Or is tobacco still just a research chassis? Has anyone moved the pathway into microbes/yeast? Has anyone actually gotten closer to microbial production of etoposide? help me with the scale production

Bibliography:

Decembrino, D., Ricklefs, E., Wohlgemuth, S., Girhard, M., Schullehner, K., Jach, G., & Urlacher, V. B. (2020). Assembly of Plant Enzymes in E. coli for the Production of the Valuable (−)-Podophyllotoxin Precursor (−)-Pluviatolide. ACS Synthetic Biology, 9(11), 3091–3103. https://doi.org/10.1021/acssynbio.0c00354
Decembrino, D., Raffaele, A., Knöfel, R., Girhard, M., & Urlacher, V. B. (2021). Synthesis of (−)−deoxypodophyllotoxin and (−)−epipodophyllotoxin via a multi-enzyme cascade in E. coli. Microbial Cell Factories, 20(1), 183. https://doi.org/10.1186/s12934-021-01673-5
Kuo, H., Wei, Z., Lu, P., Huang, P., & Lee, K. (2014). Bioconversion of Pinoresinol into Matairesinol by Use of Recombinant Escherichia coli. Applied and Environmental Microbiology, 80(9), 2687–2692. https://doi.org/10.1128/aem.03397-13
Lau, W., & Sattely, E. S. (2015). Six enzymes from mayapple that complete the biosynthetic pathway to the etoposide aglycone. Science, 349(6253), 1224–1228. https://doi.org/10.1126/science.aac7202
American Cancer Society. (2025). Cancer facts & figures 2025. American Cancer Society report
Conservation status, anticancer compounds and pharmacological Aspects of Royle: a review. (2022). Indian Journal of Ecology. https://doi.org/10.55362/ije/2022/3637