Erika Update #4: How much "proof" is needed for a proof-of-concept?
Erika Update #4: How much "proof" is needed for a proof-of-concept?
Finicky results are effectively irreproducible in life science
My PhD project was inspired by a paper from Jason Chin’s lab where they evolved the ribosome, called “RiboQ”, to work better with four-base codons. I was enchanted by the idea that we could manipulate something as fundamental as the codon size by isolating and altering the biomolecule responsible for that parameter. I wanted to apply higher-throughput engineering tools to take this idea even farther.
Unfortunately, RiboQ didn’t work in my hands. As I show in Erika Update #4 (linked below), RiboQ doesn’t seem to be any better than the normal ribosome at decoding quadruplet codons when I applied it to my system. This was the second time in four months that a published result I planned to rely on mysteriously failed to replicate (I wrote about the first of these in Erika Update #2).
After seeing this data I went back into the literature and looked at every single paper that cited the original RiboQ publication. They are all other Chin lab papers or review articles. I couldn’t find a single person who had ever actually corroborated the finding that RiboQ is better at translating quadruplet codons. Like the last time I was unable to replicate something, I still don’t think the paper is fraudulent. I think the result is probably strain/codon/media/detail specific. But practically speaking - there are enough variables, and enough friction surrounding executing experiments - that finicky results are effectively irreproducible.
As it turned out, it was alright that the published ribosome technology didn’t work the way I expected. I thought I would spend most of my time engineering the ribosome, but in reality that wasn’t the biggest bottleneck. I ended up spending most of my time engineering tRNAs and AARSs, and didn’t need to touch the ribosome. Projects don’t always go the way you think.
Erika Update #4 - Leveling up to the Ribosome
As a quick recap, this post is part of the “Erika Updates” series, where I post informal research updates I wrote in grad school accompanied by blog posts with commentary. I’m revisiting these write ups as a jumping off point for reflecting on the PhD experience, the current state of research science, and the ways we share and communicate technical content to peers and the public.
Last time I successfully tied the ability to translate four-base codons to bacteriophage reproduction. In this update, I’m leveling up and adding in a new level of complexity: the ribosome!
Here’s Erika Update #4 - 2018 3 29 - orthogonal luminescence test - RiboQ test
mRNA transcripts in cells are routed to the ribosome through Watson-Crick base parking between the Ribosome Binding Site (RBS) and ribosomal RNA Here, I’m using a very cool technique from the literature where I use a orthogonal RBS/ anti-RBS pair to route my transcripts to a separate population of ribosomes that I can engineer.
In principle, this was going to be the basis for engineering the ribosomes themselves to decode a four-base-codon genetic code. In practice, engineered ribosomes from the literature don’t work the way it says on the package, and using them places a huge burden on the cell that makes these bacteria tough to work with.
Where’d it end up?
The data about the orthogonal ribosome and RiboQ both ended up as Supplementary Figure 9 of Paper #1. The other data as usual was re-measured later using a more sophisticated data collection technique.
Behind the scenes tips
I’ve met Jason Chin. He’s delightful and very creative and I love him. He is such an amazing thought leader in genetic code expansion, and has published a ton of really great ideas. However, like RiboQ, Chin lab papers are usually publishing the first proof of concept of a new technique, and it doesn’t always work super well or reliably. Here’s a few other examples of Chin-lab tech that other scientists have improved upon. If you want to work with these, don’t start with the Chin lab version, start with v2 that has been further improved by someone else:
The original orthogonal RBSs are not very orthogonal. Ahmed and Fan had to re-engineer them so they have lower crosstalk in their orthogonal ribosome paper.
The Chin Lab’s 2021 batch of orthogonal AARSs are also not very orthogonal. The minute you try to evolve them they will promiscuously charge canonical amino acids.
The Chin lab stapled ribosomes don’t work well either, although Mike Jewett’s tethered ribosomes seem to be better (same concept, different implementation).
Want more?
If you want to follow along with this project, you can get updates by signing up through substack below, or following me on linkedin or twitter.