357 | Jeff Coller on mRNA, Vaccines, and Bespoke Therapeutics
What this covers
Sean Carroll talks with Jeff Coller, an RNA biologist at Johns Hopkins, and the conversation moves from basic molecular biology to the clinic. It begins with how the ribosome reads the genetic code, why codon assignments are largely arbitrary (Crick's Frozen Accident), and why an mRNA that is read quickly lasts longer than one that is read slowly. It then turns to origins. Most molecular biologists now favor the RNA world hypothesis, helped by Cech and Altman's finding that RNA can catalyze reactions. Coller stresses that how chemical building blocks become self-perpetuating life is still not understood.
The medical half starts with COVID. Coller notes that mRNA had over 100 human trials before 2019, and that Moderna could swap the SARS-CoV-2 spike sequence into an existing influenza program. He describes why the molecules need lipid nanoparticles, why they degrade within hours, and why they never reach the nucleus. He also covers cancer: Jim Allison's principle of training the immune system on tumor-specific proteins, and a small Sloan Kettering pancreatic cancer trial with striking survival figures.
The most concrete case is KJ Muldoon, an infant with CPS1 deficiency treated with a personalized base editing therapy delivered as mRNA, who no longer needs a liver transplant but is not cured. Coller connects this to the roughly 7,000 genetic diseases and to the difficulty of reaching organs like the lung through its mucus. The episode closes on a contested forecast: AI-designed bacteriophages suggest the same tools could be used to design dangerous pathogens.
Coller argues that mRNA is a programmable, fast, cheap instruction set for the body, and that its main limits now are delivery to target tissues and a drug regulatory system built for blockbuster drugs rather than bespoke therapies.
- mRNA can be designed in silico and manufactured in vitro far faster and cheaper than protein vaccines, which is why it worked for COVID-19
- Delivery beyond the injection site and liver, including the brain and lungs, is the main technical bottleneck
- The FDA framework economically fails for ultra-rare, one-patient therapies such as the KJ Muldoon base-editing case
Easier design lowers barriers for both rare-disease therapies and harmful viruses alike.
- Sean noted that cancer and rare-disease therapies may be cheap to design with mRNA because the same design speed used for COVID vaccines applies, so a therapy for a disease affecting only 12 people could now be worth developing.
“we can relatively inexpensively create new therapies for diseases that not too many people have”
- As it becomes easier to design harmful viruses, countermeasure efficacy improves simultaneously, so bioweapons and bioprotection develop in parallel, and the resources of a nation-state are not needed to start the design race.
“there could be, I guess, bioweapons and also bioprotections improving in their efficacy simultaneously”
Using RNA for flexibility and DNA for storage is a more sophisticated system.
- Sean suggested that RNA may be a good first try because it is flexible and can do many things, but DNA's greater stability makes it better for long-term storage, so using both is a more sophisticated system.
“RNA is sort of more flexible and can do more things but is a little bit less permanent, so it might be the good first try”