How to build a better vaccine from the comfort of your own web browser
Hannah Wayment-Steele, originally published on Eterna.
The bad news: The SARS-CoV-2 pandemic will not end until the majority of the world’s population gets vaccinated. And, there’s a problem with some of the forerunning vaccine technologies — the molecules they’re made of are prone to degrade.
The good news: We predict that the stability of any RNA vaccine can be improved at least two-fold through design. Participants in the online RNA design project Eterna showed us this was possible.
We’ve shared our predictions in a recent preprint, and we’re now in the process of testing and validating these predictions. There’s ongoing work where we need your help and creativity to make RNA vaccines even better!
How do RNA vaccines work?
For the genetic instructions in our DNA to be carried out, the instructions are first rewritten into messenger RNA molecules, called mRNAs for short. (In this article, we’ll be showing DNA as purple, RNA as orange, and proteins as green.)
The message in an mRNA is a series of codons — triplets of nucleotides that each specify a particular amino acid. A molecular machine called the ribosome reads this code and uses it to build the protein. Often, several codons will encode for any amino acid. So even though one mRNA specifies one protein to be coded, any one protein has many different mRNAs that will produce it. (We’ll come back to this later!) For instance, hemoglobin is a protein your body is using all the time — it carries oxygen around in your red blood cells. You have cells (hopefully) making hemoglobin right now.
Instead of encoding a protein like hemoglobin, mRNA vaccines code for a protein that’s not normally in our bodies: the spike protein from the SARS-CoV-2. The mRNA is injected, our ribosomes create the spike protein, and our immune systems learn to recognize it, preparing for the real virus to come along. RNA vaccines have a lot of advantages over current vaccines: they are much faster and cheaper to make, and they show promise so far in publications and clinical trials.
What’s the problem with RNA vaccines?
There’s a problem with using RNA as a medicine — it’s an unstable molecule. Its chemical composition means that over time, the strand of RNA is going to degrade. (This is why researchers think DNA, its more stable cousin, ended up being the material of our genetic code, and not RNA). Heat, UV, and other compounds in vaccine formulations can speed this up. This is a major problem if we want to deliver vaccines to developing countries, where a refrigerated supply chain may not be possible.
What can we do to stabilize RNA medicines?
RNA forms base pairs, just like DNA does — instead of G-C pairs and A-T pairs in DNA, RNA has G-C and A-U pairs. Base-paired RNA is less prone to degrade than single-stranded RNA. We developed a model to predict the stability of a molecule based on how base-paired it is.
Below is an example mRNA for a tiny protein, where each base is colored by the probability that it’s unpaired. We see that the molecule has “hot spots” in yellow where it’s more likely to degrade.
So, the question that we set out to answer is, how stable can we make a messenger RNA?
We score molecules based on the Average Unpaired Probability (AUP) over all the bases. The p(unpaired) for each nucleotide is predicted by algorithms designed for this purpose and used widely in RNA structure prediction. A lower AUP means that the molecule is predicted to be more stable.
Remember — for any protein, there are many mRNA sequences that code for it. We can write down every single one if the protein is small enough.
The RNA molecule above codes for a protein with the sequence MDYKDDDK. That’s the FLAG Tag protein, a short protein used commonly for purifying proteins. It turns out that the sequence above is the mRNA for the FLAG Tag that has the lowest AUP possible.
But, for longer mRNAs, we can’t enumerate all the possibilities. For instance, for one RNA vaccine based on the SARS-CoV-2 spike protein we designed, there’s
mRNA sequences that code for it!
For longer mRNA molecules, how low can we make AUP? That is, how much can we stabilize a vaccine molecule through designing the sequence?
We asked participants of the online RNA design project to see how low they could take AUP for some example vaccines for the SARS-CoV-2 virus, as well as some proteins that are commonly used to study protein production. We compared their designs to sequences from other common methods, such as randomly selecting codons, commercial algorithms that aim to optimize the codons that are used, as well as algorithms that are able to increase the number of base pairs.
Below: on the left is an example mRNA from a standard commercial method to design an RNA. On the right is a stabilized mRNA (this one happens to be from an algorithm we developed, named RiboTree).
We found that for five different kinds of puzzles, we reliably found a two-fold decrease in AUP — which translates to a two-fold increase in half-life!
We found that Eterna players could get a wide variety of structures, too! This is important because we don’t know precisely what sorts of structure features will help or hinder the vaccines in other parts of biology. But Eterna players are creating the molecules that will help us answer these questions.
How can I get involved?
There’s still a lot we don’t understand about how to stabilize RNA that we’re trying to figure out on the Eterna platform.
For instance, AUP doesn’t account for sequence in predicting degradation — it assumes that if an “A” in a loop is unpaired 50% of the time, or if a “U” in a loop is unpaired 50% of the time, they degrade at the same rate. But experimental evidence suggests that this doesn’t always hold.
We’re collecting more experimental data in high-throughput on shapes that Eterna players are designing! Below you can see that these predictions are largely true, but there’s a lot of discrepancies.
We need as many minds on board as possible to create new sequences that will explore the diversity of RNA sequence space as creatively we can, and sets of eyes on new data that are coming out to find patterns of how RNAs degrade.
With your help, we can invent medicines that will end this pandemic. Join us today!
Text & illustrations by Hannah Wayment-Steele
