Notes from the Lab: I Killed GFP

TLDR: Even bulletproof proteins can fail spectacularly with the wrong mRNA sequence.

‍

The Scene of the Crime

GFP has been the reliable control in molecular biology for decades. When you need to check if your expression system is working, you reach for green fluorescent protein—so dependable that its failure usually means you've made a pipetting error.

Which is why I was particularly intrigued when our GFP refused to glow.

‍

I had successfully killed GFP.

‍

An Experiment in Controlled Failure

We've been developing tools that optimize mRNA sequences across multiple axes: expression, stability, translation efficiency, and various other factors that influence potency. But here's the thing about optimization tools—they're equally good at making things worse if you point them in the wrong direction.

So naturally, I had to try it. 😀

I designed three GFP variants engineered for failure:

• The Knot: Creating secondary structure so dense that ribosomes would need a machete to get through
• The Time Bomb: Sequences littered with hydrolysis “hot spots,” essentially pre-programmed for self-destruction
• The Anti-Babel Fish: Deliberately poor codon choices—using the rarest codons in patterns that would make the ribosome weep

Same protein sequence. Radically different nucleotide sequences. The results? All over the place. Highly expressing sequences rapidly degrading. Ultra-stable sequences that hardly expressed. Six variants couldn't even be assembled despite multiple attempts.

I had successfully killed GFP.

‍

Why This Should Worry You

Here's the uncomfortable truth: every mRNA therapeutic in development exists somewhere on a performance spectrum. The exact same protein can exhibit expression levels that vary by orders of magnitude depending solely on how you encode it at the nucleotide level!

‍

‍

Most researchers don't realize they're essentially rolling dice. When you order an mRNA from a service provider or design one with standard tools, you're getting a random draw from a performance distribution. 

Maybe you land near optimal! Maybe you’re drawing from the wrong distribution? Maybe you get our dead GFP.

‍

Every mRNA sequence is designed…either intentionally or by accident.

‍

The Hidden Variable in Your Experiments

This creates a particularly insidious problem for screening studies. You're comparing Protein A to Protein B. Protein A performs better. Conclusion: Protein A is superior.

‍But is it? Or did Protein A simply win the sequence lottery while Protein B got stuck with a suboptimal nucleotide sequence?

Think of it as a phase diagram. There's a region where everything aligns: sufficient stability for manufacturing and delivery, optimal structure for translation, appropriate codon usage for your target cell type. But without intentional sequence design, you can't distinguish between protein performance and sequence performance. You might be discarding your best therapeutic candidate because it happened to get a poor nucleotide sequence.

If our deliberate effort can kill GFP—one of the most robust proteins in biology—what's happening to your therapeutic proteins? How many clinical failures stem not from bad biology but from bad sequence design?

Every mRNA sequence represents a design choice, whether intentional or accidental. The question isn't whether sequence design matters. It's whether you're making these choices consciously or leaving them to chance.

‍

What's Next for Us

We're systematically mapping the sequence-structure-function landscape across different proteins and cell types. Each "failed" sequence teaches us something new about the grammar of mRNA design. Our goal isn't just to avoid the disasters (though, that's certainly nice). It's to find sequences that allow proteins to reach their full therapeutic potential—stable enough for delivery, structured for optimal translation, and tuned for specific biological contexts.

Because if we can kill GFP on purpose, imagine what we can achieve when we're trying to help proteins thrive.

When you're ready to optimize your lead candidate for early-stage screening, we can help ensure you're making decisions based on biology, not chance. Drop us a line to discuss your specific challenges.