https://www.science.org/content/blog-post/cranking-protein-expression-limit

I found this to be an interesting thought experiment that’s been brought into reality. In chemical biology and protein research we spend a lot of time getting cells to express particular proteins for us (and indeed this is a very important commercial process for protein-based pharmaceuticals). It’s a bit of an art form. We know some general principles of how to get this to work - particular types of cells, promoters to use next to the sequence of your protein to make it get expressed more strongly, where in the cellular genomes you’d want to insert these constructs, and so on. But these don’t always work (to put it lightly) and even when they do there’s often tweaking needed to get high expression levels of properly folded proteins. Get some factors wrong, all the way down to what sort of vessels you grow the cells in, and you can end up with very low levels of those desired proteins, or perhaps very respectable levels of misfolded junk. Some readers have surely experienced both of these outcomes, perhaps even with the same damn protein.

And there’s always a limit in how hard you can press those engineered cells. You are, after all, forcing them to use their metabolic energy to produce something that they don’t want and don’t need. The key is to maximize that without killing them off, and that level will vary widely depending on the protein and the cell line. The paper linked above is trying to ask what happens when you get cells to produce higher and higher amounts of the least intrinsically cytotoxic protein that they could find to see what the stress pathways really are.

The authors settled on a fluorescent protein that was mutated not to fluoresce (mox-YG) and a glycolytic enzyme that was mutated to be nonfunctional (Gpm1-CCmut). These could be revved up to rather high levels without doing anything on their own (other than hogging cellular resources and physical space). That really is one of the recognized categories of trouble at high expression levels, “resource overload”. That is, the systems responsible for protein production are so tilted towards making the foreign protein that production of essential proteins for the cell start to be disrupted. Then there’s “stoichiometry imbalance”, especially found when you’re expressing a whole protein complex, as well as “pathway modulation”, and “promiscuous interaction”. These can overlap a bit or be linked together, but involve the large amounts of foreign protein interacting with existing cellular proteins to the detriment of their natural functions. The model is that the expression limit hits one of these barriers, and if you find a way to remove that, then it will go up until it hits the next one, and so on.

The Final Boss of this process might well be the restraints on protein synthesis itself. That’s a massive ongoing general process, whereas particular transport or degradation pathways have more specific subtrates and functions. If you’re overloading the capacity of the ribosomes and transfer RNA pathways, you’ve pretty much pegged the system on the far right side of the meter. (The manuscript has a number of references to studies over the years investigating these).

In yeast cells, things seem to max out at about 15% of the total protein concentration of a cell, which you have to admit is rather a lot for a single protein to be taking up. But no one has been sure that this is the real limit. The authors here note that some proteins have more efficient mRNA pathways than others, for example, putting a burden on transcription. If that’s well-lubricated, then you might run into limits on translation! And so on. Where do the bolts start to come loose?

Well, this paper got the levels of that formerly fluorescent mox-YG protein up to over 40% of total protein in the cell, which certainly shatters the old record. And at those levels the limiting factors seemed to be outright amino acid depletion, problems with ribosome expression, and an apparent metabolic switch from glycolysis over to more oxygen respiration. But they also got their other “benign” protein (Gpm1-CCmut) up to these levels without seeing those signs of nitrogen starvation or increased oxygen usage (!)

So this is certainly progress, but the fundamental questions remain open. The authors note that it will take a better and more continuously “tunable” expression system to learn more (for example, at what point does that respiratory switch start to kick in?) And we still need to understand more about how the effects of overexpression, even at these severe levels, can still be so distinct. There’s a lot that can go wrong!

https://www.science.org/content/blog-post/cranking-protein-expression-limit