How do you make a new protein, or a new function in an existing one?
This is the goal of the field of protein engineering. Researchers working in this field use a number of strategies to try to make proteins with new characteristics. The development of proteins with new function have applications in industry, medicine, and biotechnology.
Want a more stable or more efficient enzyme? Talk to a protein engineer.
Want to convert a protein with one function to a completely new activity? talk to a protein engineer.
Want to make a molecule with a function never before seen in nature? talk to a protein engineer.
I was able to hear from many working in the field this weekend at the second iteration of the Protein Engineering Canada meeting, held at the University of Ottawa. The meeting was brief but full of excellent talks. There were some common principles that kept coming up that I’ll try and summarize here.
Reductionism fails in engineering proteins (for now)
Elan Eisenmesser gave an excellent talk about protein dynamics and mentioned how “the worst thing to happen to biochemistry was biochemists”. The argument is that biochemists like simple, reductionist models, but as we’ve studied proteins and how they work, we know the truth remains much more complicated than that. Many times to change the function of a protein, changes far from the site of interest are necessary, and we remain pretty terrible at predicting what those might be.
In many talks at this meeting it came up that the most effective modified enzymes are typically achieved through non-intuitive mutations. So, if we limit ourselves to deterministic changes where we predict the results, we may miss most possible opportunities to develop new properties in a molecule. Rational approaches can lead us the wrong way – screening of many different sequences are necessary to find proteins with desired properties. We may someday be able to predict function from sequences alone, but those days remain far in the future.
The challenges in screening and sorting for function
The number of possible sequences even in a relatively small protein are astronomically large, 20 residues raised to the power of the number of residues in the protein. As a result, it’s never possible to screen for function in every possible sequence. It is always necessary to reduce the number of testable sequences and structures of proteins to test to a manageable level, and this needs to be done in a smart way.
Two strategies of screening a restricted library presented were Tim Whitehead’s group’s strategy of systematically replacing every amino acid in a protein with every possibility and compete the bacteria against each other to try to alter function. Another method was Justin Siegel’s to look to nature and the diversity of sequences in the environment to find better proteins that have developed in the wild. These strategies guide us to finding new functions without having to individually screen 20100 individual proteins, something that we would still be doing until the death of the universe.
A third strategy discussed was the generation of a collection of completely new proteins, never seen before, to screen for brand new functions in proteins:
Making something from nothing
It’s much easier to break something than to build it new from scratch. This is a fact of life, dictated from the rules of thermodynamics.
But, at the same time, it’s surprisingly easy to build something. Michael Hecht gave an excellent talk indicating how his group has made a library of brand new proteins, and screens them for function. He described that for some functions, many new proteins could carry them out, without any evolution to tune the protein’s activities. So, if you have the right type of function, it might be possible to find that function in randomly-generated (although constrained) protein sequences. This has some pretty profound implications for understanding the origin of life as we know it.
The finding from this work, and in the analysis of a large family of enzymes presented by Janine Copp was that you can relatively easily get weak, promiscuous activity from primitive enzymes, which then are refined to more specialized proteins with higher specificity and activity. This is the case in nature, and also now in the lab where researchers develop new and better functions in proteins by driving them to specialization.
Getting comfortable with disorder and dynamism
Similar to problems with reductionism, an assumption that x-ray crystal structures have convinced many of is that proteins are mostly rigid and don’t move much as they carry out their function. This isn’t really true, I’ve ranted before on this site about why we need to undersand molecules are jiggly. Proteins are nearly chaotic, with interconnected networks of interactions that drive their function.
A recurring theme in this meeting was that an understanding of dynamics is necessary to develop a good grasp of function and how to change it in a molecule. Many talks including Sophie Gobeil’s talk on an antibiotic resistance enzyme and Adam Damry’s award-winning talk on engineering of a dynamic function in a protein touched on these points. It is necessary to understand dynamics to predict function, and while this remains challenging, it is possible to develop some predictive insights through carefully constructed experiments.
The emerging art of protein design is starting to mature, guided by a more comprehensive understanding of protein function, and smart strategies of how to get there. I’m excited to see where the field is headed!
A meeting well spent
In addition to all this work on engineering of proteins, my jaw dropped to see some of the amazing new T3SS structures coming out of Natalie Strynadka’s group, and Martin Schmeing’s presentation of his group’s megaenzyme studies, appropriately set to Miley Cyrus and Taylor Swift.
Overall, an amazing meeting, I hope to go again in 2018. If you’re interested in protein design and engineering, I can’t recommend the meeting enough. Hope to see you there.