See the nav bar on the left to look through my work for How To Biomanufacture Almost Anything (Fall 2026)!
(1) What you would like to biomanufacture and why?
I want to make therapeutic exosomes from stem cells to help the heart heal after a heart attack. So what’s an exosome? Cells are constantly sending messages to each other. One way they do that is by packing molecules into tiny bubbles called exosomes and releasing them. They’re really small, about 30 to 150 nm across. Inside, they carry microRNAs and proteins. When a stem cell’s exosomes reach damaged heart tissue, that cargo can calm down inflammation, cut down on scarring, keep the surviving heart muscle alive, and help new blood vessels grow (McMullan et al., 2025).
I got to this idea through my own research on cardiomyocyte differentiation. For years, the plan for fixing a damaged heart was to transplant stem cells or new heart cells into it. The problem is that most of those cells die within a few days. What surprised researchers is that the heart still got a little better anyway. That suggested the cells weren’t helping by turning into new heart muscle. They were helping through the stuff they released before they died (Joladarashi & Kishore, 2022). So my thinking is: why transplant fragile cells at all? Why not just make the part that’s actually doing the work, the exosomes?
Going this route has a few big advantages over transplanting cells:
It’s safer. No living cells go into the patient, so there’s no risk of transplanted cells causing irregular heartbeats or turning into tumors. The immune system is less likely to reject it. Exosomes are much less likely to set off an immune response than whole donor cells.
It can be ready ahead of time. Exosomes can be frozen and stored. A hospital could keep doses on the shelf and use them as soon as a patient comes in. With cell therapy, the weeks-long manufacturing process would only start after the heart attack already happened.
Who it’s for: people recovering from a heart attack or with epopel who have damaged heart tissue. They’d get it in the days after doctors open the blocked artery. The goal is to save as much heart muscle as possible before it’s permanently lost and replaced by scar tissue.
(2) What living system would you use?
Chassis: I’d use a mammalian system: human mesenchymal stromal cells (MSCs) grown in a bioreactor. MSCs are adult stem cells that naturally release exosomes, and they have the most research behind them for heart repair (Joladarashi & Kishore, 2022). I picked a mammalian system over a microbial or cell-free one because bacteria and yeast can’t make human exosomes. The product has to come from human cells, since the exosome cargo is the medicine. My backup chassis is iPSC-derived cardiac progenitor cells, which are closer to my own research but harder to grow at large scale.
How it works:
What goes in: MSCs, plus nutrient media with glucose, amino acids and growth factors, and controlled oxygen. One key detail is that the media can’t contain normal fetal bovine serum, because that serum is full of cow exosomes that would contaminate my product. So I’d use serum-free media.
What biological process happens: First the cells divide until the bioreactor is packed with them. Then I lower the oxygen for about 2 days. This mimics the low-oxygen environment of a heart attack, and studies suggest it makes the cells load their exosomes with more heart protective microRNAs. The whole time, the cells are naturally packaging microRNAs and proteins into exosomes and releasing them into the liquid around them.
What comes out: liquid media full of exosomes. I’d concentrate it with tangential flow filtration, then purify the exosomes by size with size exclusion chromatography. The purified exosomes get frozen and stored until a patient needs them.
My assumptions:
Each cell releases about 2,000 exosomes per day. Published numbers vary a lot, so this is a middle-low estimate. The bioreactor reaches 1 × 10e^7 cells/mL. About 40% of exosomes survive purification.
(3) How would you use automation?
I’d combine my cells with two automation technologies.
An automated perfusion bioreactor (automating culturing and sampling). This bioreactor continuously pumps in fresh media and removes used media, and its sensors track oxygen, pH, glucose and cell density in real time. Why: with exosomes, the cells’ condition is the product. If glucose runs low or the pH drifts, the cells get stressed and change what they put in their exosomes. Then one batch could work and the next might not. Automation keeps conditions the same every time, and it makes the low-oxygen step exact instead of someone adjusting it by hand.
An Opentrons liquid-handling robot (automating measurement and screening). Every batch has to be tested before it’s released. That means counting particles, checking for the exosome surface markers CD9, CD63 and CD81, and running potency tests. All of these need tons of precise dilutions, and small pipetting mistakes throw off the counts. Why: the robot does this faster and more consistently than a person. It also lets me test many samples at once during prototyping, like comparing different oxygen levels or harvest days. This fits the point in Petzold & Mukhopadhyay (2025) that automation matters most in the testing and optimization loop.
(4) How much would you need to make, and how long would it take?
(a) How much?
Scale: 1,000 patients. That’s roughly how many heart attack patients a large regional heart center treats in a year, so it’s a realistic first- year supply for one hospital. Dose (assumption): 5 × 10e^10 exosomes per patient. No human dose has been set yet, so I estimated this from large-animal studies.
(b) How long?
Grow cells to full density: 12 to 14 days Low-oxygen conditioning: 2 days Harvest (3 days for a safety margin): 3 days Purification: 4 days Formulate and freeze: 2 days Quality testing: 10 days Total: about 5 weeks
What don’t you know yet?
How do I know if a batch actually works? (biggest unknown) I can count exosomes, but a count doesn’t tell me whether they heal a heart, and there’s no agreed potency test. What I’d need to do: test several lab assays, like how well the exosomes help grow blood vessels or protect stressed heart cells. Then I’d compare the results against real outcomes in animal heart attack models and keep the test that best predicts which batches work.
Is every batch the same? (bottleneck) Exosome cargo could change as the cells age during a run or between donors. What I’d need to measure: sequence the microRNAs and measure the proteins in exosomes from different days and batches, to see how much they drift and how long I can run the cells before quality drops.
Are my key numbers right? (biggest assumption) My whole calculation depends on 2,000 exosomes per cell per day, and published values range more than 10-fold. If the real number is 200, harvest goes from 1.25 days to 12.5 days. The human dose is also just an estimate. What I’d need to determine: measure the actual exosome output of my cells in my bioreactor, and eventually get dose data from clinical trials.
AI
Client: Claude
Prompts: I used Claude to brainstorm options, compare them, and polish the writing and formatting. I wrote and checked the final choices and numbers myself and can explain every step. “Give me ideas for a biomanufacturing proposal beyond gene editing that fit the HTBAA Homework #1 questions” “Is this possible” “Is this feasible with a limited supply and who could this be marketed for” “How would these cells be alive for more than a day”
Readings:
Joladarashi, D., & Kishore, R. (2022). Mesenchymal stromal cell exosomes in cardiac repair. Current Cardiology Reports, 24, 405–417. https://doi.org/10.1007/s11886-022-01660-1 McMullan, E., Joladarashi, D., & Kishore, R. (2025). Unpacking exosomes: A therapeutic frontier for cardiac repair. Current Cardiology Reports, 27, 73. https://doi.org/10.1007/s11886-025-02225-8 Petzold, C. J., & Mukhopadhyay, A. (2025). From bench to biofactory: High-throughput technologies and automated workflows to accelerate biomanufacturing. Current Opinion in Biotechnology, 94, 103320. https://doi.org/10.1016/j.copbio.2025.103320