Building a better protein in the hope of better therapies

Dr. Phil Bradley remembers with crystal clarity seeing the image that would jumpstart a new path in his research career.

The Fred Hutchinson Cancer Research Center computational biologist was looking at the latest results of a computer program he’d written to model possible engineering tweaks to a natural protein, tweaks that could allow researchers to use the protein in new medical applications.

In nature, these particular proteins wrap themselves around DNA in a very precise way, Bradley said. But one of the models he’d designed turned out to have what was basically a bug in the program — it closed on itself instead of around the DNA.

And it was gorgeous, Bradley said.

“There was this structure of these [helical shapes] outside and inside and kind of like this,” he said, describing the modelled protein’s barrel-like shape with animated gestures, “and they were coming around and forming this beautiful closed shape.

“I wanted to make that thing right away.”

That moment launched Bradley’s first foray into de novo protein design — a brand-new field in which scientists design and create proteins unlike anything seen in nature.

In a study published Wednesday in the journal Nature, Bradley and his colleagues at the Hutch and the University of Washington’s Institute for Protein Design debut the latest proteins that have come into this world fully engineered by people, rather than by nature: tiny, donut-shaped proteins made of simple but unnatural repeated segments.

It’s the very early days for this kind of protein engineering, said Fred Hutch structural biologist Dr. Barry Stoddard and co-author on the study.

“The number of successes in this field — you can still count them pretty easily,” Stoddard said.

But he and other designers of novel proteins have big hopes for their tiny creations. Lab-made proteins could be stepping stones to advances ranging from better therapies for a variety of diseases to entirely new nanomaterials built from protein bits.

Right now, much of de novo protein design falls under the category of “let’s see what we can do,” Bradley said.

His hopes for the future of his mini-donut creations are founded on the myriad tasks proteins accomplish in the natural world. The workhorses of the cell, proteins carry out nearly every function in every living creature.

“Protein design is conceptually really wide open … When you look at the structures of naturally occurring proteins, it’s an amazing universe of different shapes and forms,” Bradley said. “It’s really cool what natural proteins do. To be able to recreate some of that functionality using computers would be pretty fantastic.”

Artificial donuts for your health

The new inventions are different from actual donuts in a few key ways, Bradley said. They’re fat- and sugar-free, for one. But more importantly for the mini-donuts’ potential for clinical use, they’re remarkably symmetrical and are able to keep their shape under a variety of harsh conditions.

As a former mathematician, Bradley likes the proteins in part for their regular, mirror-image aesthetics. But their pleasingly round forms also make them highly predictable, which, in theory, could translate to some intriguing medical applications. Bradley and Stoddard believe therapies built on the back of the donut proteins could one day yield more potent vaccines, more powerful growth factors to stimulate the growth of cancer-fighting immune cells in stem cell or cord blood transplants, or even a molecular sponge to soak up toxins in the body after poisoning.

Think antibodies, growth factors and some other hormones, certain vaccines — all proteins. Such therapies generally work by sticking to something else in the body, often to another protein on a cell’s surface. The researchers want to test the engineered donut proteins as mini-scaffolds housing multiple, regularly spaced copies of a given protein therapeutic. The more copies of the therapy you can bring into close contact with the target cell, the better, the theory goes.

“Proteins are now increasingly being used as therapeutics,” Stoddard said. He thinks the donut proteins could be a simple addition to multiply those therapies’ power: “We’re very interested in what happens if you create a molecule where multiple copies of a protein therapeutic are presented simultaneously.”

Stoddard and Bradley are now working with Fred Hutch immunotherapy researcher Dr. Stan Riddell and vaccine expert Dr. Larry Corey in an early-stage project using their donut proteins as scaffolds for medical therapies. They’re gearing up for experiments to test whether presenting multiple copies of a growth factor or a cancer vaccine will improve those treatments’ abilities in the lab.

The protein design pipeline

The donuts themselves look simple enough, but de novo protein design is no easy task. It requires not only a deep knowledge of the 3-D shapes of natural proteins, but the ability to predict those shapes based solely on their sequences of amino acids, proteins’ building blocks.

Bradley has spent much of his research career honing such knowledge. As a postdoctoral fellow working with UW biochemist Dr. David Baker (also a co-author of the Nature paper and a major force in the protein design world), Bradley played a key role in the development of Rosetta, software that automatically predicts how proteins “fold” based on their sequence.

Still, despite Bradley’s years spent studying the minutiae of proteins, creating them from scratch turned out to be more challenging than he anticipated.

The team’s protein design pipeline starts with a computer program Bradley wrote to generate sequences that will yield the 3-D shape they’re looking for. Once the modelled proteins look good on the computer screen, Bradley passes his sequences to Stoddard lab research technicians (and co-authors on the study) Lindsey Doyle and Jazmine Hallinan, who use engineered bacteria to create the physical proteins. Finally, Stoddard’s team purifies the designed proteins and captures their 3-D structure using a painstaking and often finicky technique known as protein crystallization.

And then they compare those 3-D pictures to the original computer design to see how close they came, making tweaks to the computer program in later rounds as needed.

Their first try was pretty good, Bradley said. Out of the three proteins they tested, one crystallized right away and looked just like his initial design. But of the 10 proteins he designed in the second round, only one even yielded crystals at the final step in the pipeline, meaning the other nine couldn’t be verified.

“Then the honeymoon was over. It became clear that things were actually pretty hard,” Bradley said. But he was already engaged, and their third iteration yielded several successful proteins. “Now seeing how much fun it is I’m hooked,” he said.

Bradley came up with one way to keep the team motivated — donuts, the regular version. Bradley would bring a box over to Stoddard’s lab after every successful step in the project.

“We get crystals and then the box of donuts would appear,” Stoddard said. “We like donuts — both molecular and the edible kind.”