Week 1 HW: Final Project Ideas

cover image cover image

Homework #1

  1. What I want to biomanufacture, and why I would like to work in researching and making something along the lines of engineered human cells that get injected next to a painful joint, and release an anti-inflammatory protein and are controlled through magnogenetics. One of the potential proteins I could user is IL-1Ra, sold as anakinra. My target is chronic spine/neck pain from facet joints. I have it pretty severely, and have done loads of treatments through years, which is why I picked it. About 18.7% of U.S. veterans report neck pain and 49.6% report joint pain (NHIS 2018), both well above non-veteran rates. The cervical case is interesting because the options are thin. Steroid injections wear off, the main durable treatment is radiofrequency ablation, burning the nerve that carries the pain signal. It works in someways, but the nerve grows back in six to eighteen months, and you do it again. The current plan is destroying a nerve on a recurring schedule. It’s also where local control matters most. The cervical joints are small and sit within a centimeter or two of the spinal cord, so you can’t solve the problem by flooding the body with drugs and it is too risky to intervene surgically. A magnetic field potentially fits as it passes through tissue, it’s non-invasive, it can be aimed at one joint or area, and a patient could use it on a schedule.
  2. What living system would I use? Human cells, grown in a lab and engineered so the patient’s immune system tolerates them. Inside those there would be a switch out of three parts. A receptor that works like a pull-cord. Most cell receptors are locks: the right chemical key arrives and the lock opens. Here a synNotch/SNIPR-type receptor would stick out of the cell surface folded shut around a hidden cut site. The magnet would then be used for it to unfold, exposing the site. Enzymes in the membrane snip it, the inside end falls off, drifts to the nucleus, and switches on a gene. It doesn’t respond to chemistry. Magnetic beads as the handle. The receptor is built to grab one specific molecule (GFP), so I’d glue GFP onto tiny iron beads. Now the receptor is holding something a magnet can grab. A gene for the anti-inflammatory protein, placed after the switch. Swap that gene and the same machine makes something else. In: culture medium at 37 °C, the GFP-coated magnetic beads, and a magnetic field. What happens: the magnet pulls the beads, the beads pull the receptor open, the gene turns on. Out: the anti-inflammatory protein, at a rate set by the field. Is it physically possible? There is precedent. Gordon et al. pulled on Notch1 with magnetic tweezers and ligand-coated magnetic beads, using a 96-well magnet array [2]. They found the receptor opened at about 1.4 pN in living cells, and my rough estimate is that a magnetic bead in a strong gradient supplies around 12 pN — enough, but not by much. Assumptions:
  3. The magnet really pulls properly. Everything rests on this.
  4. It shuts off in hours, not weeks. Otherwise it isn’t a switch, it’s a delay. It would probably need some degradation tag on the protein.
  5. A wearable magnet reaches the joint. Cervical facet joints sit only 2–4 cm deep.
  6. The cells and beads stay where they were injected for months. If they drift or get cleared, this becomes a therapy you re-dose a few times a year rather than implant once.
  7. How I’d use automation What needs automating isn’t any one lab step it’s the testing. The switch has several settings I don’t know anything about yet. How much of each DNA piece, how many beads, how close the magnet. Im unsure of right combination, I have to test a lot of them. I would use automation through a liquid handling robot as it can run dozens of combinations at once, each mixed the same way, and the software picks the next round based on the last. The step being automated is the liquid handling and screening, the build-and-test loop itself, not any single experiment.
  8. How much, and how long Scale: Veterans treated at VA Boston for a year.

Patients served by VA Boston, FY2025 65,506 [4] × neck pain prevalence [1] 18.7% → 12,250 × severe enough to treat (assumption) 25% → ~3,000 patients Cells per injection 10 million Total: 3 × 10¹⁰ cells a year. That’s about 77 mL of packed cells a third of a coffee cup covers every eligible veteran at VA Boston for a year. How long? From one vial of 10 million cells that’s about 12 doublings around 12 days of growing, plus a week to harvest and freeze and 2–4 weeks of safety testing. Around six weeks for a single batch, and one batch covers the year. At this scale you could even use ordinary primary cells; the immortalized line in Section 2 only becomes necessary if you scale to the whole VA system. 5. What I don’t know yet:

  1. Whether the magnet can pull hard enough. This is the assumption everything else rests on. There’s a well-known critique in this field as Meister showed that one version of magnetogenetics falls short of the energy it would need by five to ten orders of magnitude [5], and other labs couldn’t reproduce the original results [6].
  2. Whether the beads are safe enough to put near a spine. Iron particles sitting on cells aren’t harmless, and the uncomfortable possibility is that the thing delivering the signal is also the thing damaging the cells. Near the cervical spine that matters more than it would in a knee.
  3. I’m coming brand new to the field of biomanufacturing. It is going to take a lot of time and research to pursue what I am initially proposing. AI use: Claude, September 20, 2026. I inquired in building a proposal around magnetic control of engineered mammalian cells, which I first saw in a classmate’s HTGAA 2026 final project (Sami Syed, Magnetically Controlled Neuromorphic Computation) in relation to my personal struggle. I also used it to check whether the science is viable, and to explain how the biological system would function. I also used it to project calculations in testing groups in relation to the VA Boston data. References
  4. Taylor, K. A., Kapos, F. P., Sharpe, J. A., Kosinski, A. S., Rhon, D. I., & Goode, A. P. (2024). Seventeen-year national pain prevalence trends among U.S. military veterans. The Journal of Pain
  5. Gordon, W. R., et al. (2015). Mechanical allostery: evidence for a force requirement in the proteolytic activation of Notch. Developmental Cell
  6. Petzold, C. J., & Mukhopadhyay, A. (2025). From bench to biofactory: high-throughput technologies and automated workflows to accelerate biomanufacturing. Current Opinion in Biotechnology
  7. U.S. Department of Veterans Affairs. VA Boston Healthcare System, About Us (FY2025 patient volume: 65,506)
  8. Meister, M. (2016). Physical limits to magnetogenetics. eLife
  9. Xu, F.-X., et al. (2024). Magnetogenetics as a promising tool for controlling cellular signaling pathways. Journal of Nanobiotechnology, 22, 291