University of Washington biochemist David Baker at the 2015 AAAS Annual Meeting.| Atlantic Photography Boston
SAN JOSE, California — Like a molecular mechanic, David Baker has been tinkering under the hood of proteins for years to figure out how they work. So maybe it's only natural that his next desire was to build lean, mean protein machines of his own.
In his plenary address at the 2015 AAAS Annual Meeting, Baker described how he and others are designing and building proteins from scratch, practicing what he calls "post-evolutionary biology." The results are customized proteins that might block the flu virus, or tweak immune system functioning, or serve as building blocks for nanowires and cages.
"Rather than starting with protein sequences from genes that have evolved over billions of years that we see in modern organisms, the design problem is to basically make up brand new proteins that don't exist in nature, and then bring them into existence," said the University of Washington biochemist.
Baker's research focuses in part on the prediction of protein structures, or solving the puzzle of how a linear chain of amino acids folds up into the 3D structure of a protein. Toward that end, he led the development of the Rosetta computer program that predicts molecular design, and extended its reach with citizen science offshoots like Rosetta@Home and Foldit. Rosetta@Home has participants lend their unused computing power to identify possible proteins structures, while Foldit has become an online puzzle game for humans willing to figure out the folding themselves.
Baker and his colleagues turned this process on its head with a 2003 paper in Science — winner of AAAS' 2004 Newcomb Cleveland Prize — that showed how to build a protein from scratch using lessons learned from the Rosetta projects.
Building a new protein begins with a computer calculation for a new amino acid sequence that seems like it will produce a protein with a certain structure or function, Baker said. After that, the researchers back-translate the amino acid sequence into a genetic sequence, buy the DNA base pairs to build the gene spelled out by that sequence, and then turn the gene loose in bacteria or yeast that have been programmed to manufacture the protein.
"One of the interesting questions you can really get at this way is whether nature has really sampled all the different kinds of proteins that can exist, or whether there are some types of proteins that evolution never got to," said Baker.
He and his colleagues are moving much faster than evolution at the moment, and have built new structures like repeating proteins that look like flat struts, and multiple protein helices tied together with bonds more durable than those found in nature.
When they want to build a protein with a specific function, they approach the design like a climber approaches a rock wall, "to find out where the handholds and footholds are and then figure out how to place your body," Baker said. In this case, the handholds are the amino acid side chains of a protein that will interact with other proteins or small molecules, and the body is the folding of the protein that incorporates all the amino acids into one structure.
Using this technique, the scientists have designed proteins that latch on to a flu protein and weaken the virus when it infects mice and ferrets. They have built proteins that disrupt other protein interactions in cells, including the interactions that prevent cancer cells from dying.
Researchers have also created a whole kit of proteins that might be used someday to build nanowires that carry electrical signals, cages that contain and safely deliver drugs, or latticework that acts a mat on which to grow new cells to heal wounds.
There's a bit of whimsy in all this, to take natural materials and fold them in designs never envisioned by nature. But Baker suggested that the world also needs new proteins to meet new challenges that loom on the human timescale, such as the diseases of old age, dwindling energy supplies, and a warming planet.
"If we could wait a billion years, new proteins would evolve that could solve these problems beautifully," he said, "but can we jump forward a billion years and design new proteins now that would solve these problems?"