Week 1 HW: Making FGF2 for cell-culture research

Constantin Convalexius · HTBAA Fall 2026

1. What do I want to make, and why?

I’d like to produce human fibroblast growth factor 2, usually called FGF2 or basic FGF, for research labs working with stem cells. The final product would be purified, active protein that researchers can add to their cell-culture medium.

I’m interested in regenerative biology and tissue repair. For this project, I want to understand how to make one of the ingredients that this research depends on. FGF2 seems like a useful starting point: it has an established role in some stem-cell culture systems, and there are published examples of producing it in bacteria. [1, 2]

My aim would be to develop a reliable way to produce and test it, and then work out whether the process could be affordable for research use. I haven’t done the experiments yet, so the calculations below are a first estimate of what that might involve.

2. What living system would I use?

I would start with a laboratory strain of E. coli carrying DNA that encodes human FGF2. The bacteria would need nutrients, including carbon and nitrogen sources, minerals, and oxygen. They would use their transcription and translation machinery to make the protein inside the cells.

After growing the culture, I would harvest and open the cells, recover the protein, and purify it. The finished product would need checks for identity, purity, concentration, and biological activity before being divided into aliquots for use.

The overall process would be:

Grow the bacteria → produce FGF2 → harvest the cells → recover and purify the protein → test it → prepare aliquots.

Gasparian and colleagues report recovering 100 mg of purified, active FGF2 per litre of bacterial culture. That gives me a reason to start with E. coli and a useful reference for the calculation. It doesn’t mean I would immediately get the same result. [2]

Cell-free production is another option I could explore if producing or recovering FGF2 in bacteria proved difficult. For this first proposal, I would stick with E. coli because there is a relevant published production result to work from.

3. How would I use automation?

The first step I’d automate is repetitive pipetting. I would use an Opentrons Flex, with an eight-channel pipette, to prepare different test conditions and dilution series for measurements. [3]

For example, I could compare 24 production conditions in triplicate. That would use 72 wells of a 96-well plate, leaving space for controls. The actual assays might need separate plates. A plate map would help keep each sample connected to its condition and results.

I would want to measure both the amount of protein and how well it works. A condition that produces a lot of protein is only useful if that protein is active. I would therefore compare samples with a reference FGF2 preparation and suitable negative controls.

The practical benefit of the robot would be more consistent transfers and less time spent on repetitive preparation. Culturing, purification, and measurement would still need their own equipment. I would have to measure the actual time saved once the workflow was running. The automation reading also makes clear that the individual steps and their data need to work together. [4]

For larger batches, I would use a stirred bioreactor with a 5 L working volume, automated control of conditions such as pH and dissolved oxygen, and chromatography equipment for purification. The liquid handler would still be useful for sample preparation and testing.

4. How much would I need, and how long would it take?

How much?

For a first estimate, I imagined supplying 100 research labs for one year. I used these assumptions:

The concentration comes from a published study of human embryonic stem-cell culture. I used it as an example; the right concentration would depend on the cells and medium. The other values describe a hypothetical group of labs, rather than measured demand. [1]

The total medium used would be:

100 labs × 1 L/(lab·week) × 50 weeks = 5,000 L/year.

Since 100 ng/mL = 100 µg/L = 0.1 mg/L, the amount of FGF2 needed would be:

5,000 L × 0.1 mg/L = 500 mg/year.

I’d add a 20% reserve for testing, losses during aliquoting and handling, and some variation in demand:

500 mg × 1.2 = 600 mg/year.

So my production target would be 0.6 g of purified FGF2 per year.

How many batches?

For the production estimate, I would assume 50 mg of purified FGF2 recovered per litre of bacterial culture. This is half the published 100 mg/L result, but I would still need to test whether I could achieve it. [2]

At that titer, the minimum culture volume would be:

600 mg ÷ 50 mg/L = 12 L.

Each 5 L batch would produce:

5 L × 50 mg/L = 250 mg.

I would therefore need three full batches, giving 750 mg from 15 L of culture. Two batches would only give 500 mg.

The 50 mg/L assumption refers to protein recovered after purification, so purification losses are already included. The earlier 20% reserve covers what happens after that. Also, the bacterial culture volume is separate from the 5,000 L of medium used by the research labs: the purified protein would be added to that medium in small amounts.

How long?

Once I had a working strain, purification method, and assays, I would initially budget a week per batch:

These times are assumptions for planning. I would need to measure them, especially the time required for purification and the activity assay.

With one reactor and batches run one after another:

3 batches × 7 days = 21 days, or about three weeks.

This leaves out the time needed to develop the process and assumes every batch passes the quality checks. I could also spread the batches across the year instead of producing everything at once; I don’t yet know how long the finished protein would remain active in storage.

Running three reactors in parallel could shorten the schedule, but only if there were also enough capacity for purification and testing.

The recovered amount makes a big difference:

This is one reason I would measure recovery early. The estimate is also sensitive to demand: if the labs needed twice as much FGF2 per litre of medium, the annual mass requirement would double.

5. What do I still need to find out?

Does the protein work as intended? I would need to check its identity and purity, measure residual contaminants such as endotoxin, and compare its activity with a reference FGF2 preparation in a cell-based assay. Measuring protein concentration alone would not answer this.

How much can I consistently recover? I would measure soluble protein production and losses during purification, then repeat the process to see how variable the result is. I would also need to check whether conditions that work in a small culture still work in a 5 L reactor, where mixing and oxygen supply can differ.

Would the finished product stay useful and be affordable? I would test activity after storage and handling, and estimate the full cost per usable milligram. That would include purification, testing, consumables, equipment, and labour, as well as the culture itself.

My first experiment would focus on recovering a small amount of active, sufficiently pure FGF2. If that worked, I would move on to repeatability, larger batches, and cost.

References

1. Levenstein, M. E., et al. Basic FGF Support of Human Embryonic Stem Cell Self-Renewal. Article.
2. Gasparian, M. E., et al. (2009). Overexpression in Escherichia coli and purification of human fibroblast growth factor (FGF-2). Biochemistry (Moscow), 74(2), 221–225. PubMed abstract; DOI. The production figure used here is reported in the abstract.
3. Opentrons. Flex introduction and Python API tutorial.
4. 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. DOI.

AI use

I used OpenAI Codex on 22 September 2026 to help develop the topic, review course materials, find sources, draft and revise the text, and check the calculations. My prompts asked for help with the first session and homework, gathering and understanding course materials more quickly, and improving the writing.