Can protein expression be ‘solved’?

That’s the title of a great Trends in Biotechnology article led by @the align foundation’s @erika debenedictis. I loved the preprint, I am thrilled to see it published. The Align Foundation is a @schmidt futures and @griffin catalyst-funded research non-profit that is seeking to achieve “predictable biology” by hosting tournaments and hosting and sharing “living datasets backed by automated, open source methods.”

I find this review + position paper extremely useful for framing the problem of “end-to-end protein expression” (my own description of the problem). The phenomenal progress we have seen in the last five years -AlphaFold, AlphaFold2-3, Rosetta Fold, RFDiffusion, ProteinMPNN, ESM and now Boltz-1 and -2, has been focusing on the “end protein”, assuming it already exists in a well-defined, properly folded, soluble state. For recombinant proteins obtained from microbial organisms, a predictive model of soluble protein expression would be a significant advance with major benefits to protein engineering, biomanufacturing or pharmaceuticals. There is expression itself (e.g., promoters, origin of replication, tags ..) and then foldability, stability and solubility of the protein expressed in the microbial host. A predictive model could then be used to optimize the various modifiable elements (e.g., picking a promoter) to maximize soluble expression.

The authors make a great case for why this is an important question and offer a roadmap to the objective of building a predictive model, starting with how to build a robust expression dataset (or datasets, one for each organism).

I would like to suggestion one addition to the model: don’t stop at end-to-end expression and include secretion to your overarching objective. The ability of a microbial organism to not only express but also secrete a protein has a lot of additional applications. The secretion apparatus has modifiable elements (e.g., signal peptide) that could also be optimized if we had a predictive model for secretion. Let’s make it the next step!