Proteins are an impressive bunch. Starting with amino acids as their basic building blocks, these complex molecules fold into intricate 3D structures and control just about every biological process that keeps us alive.
Phil Romero wants to understand how proteins accomplish that job so that he can eventually apply their power to important problems in medicine, agriculture, chemistry and bioenergy.
“Describing how proteins perform a vast array of biological functions is tremendously challenging for two reasons,” says Romero, an assistant professor of biochemistry at the University of Wisconsin-Madison. “One, they are highly dynamic molecules that constantly change their shape; and two, their properties emerge from the collective behavior of many interacting components.”
So instead of relying on a bottom-up approach that uses the laws of physics to predict biological behavior, Romero, who has an affiliate appointment with the Department of Chemical and Biological Engineering, is betting on the top-down approach: learning how protein sequence translates into function by analyzing massive data sets.
“Our ability to generate, store and analyze biological data has exploded during the last decade,” Romero says. “That’s why data-driven approaches are playing an increasingly important role in biological discovery and engineering.”
Romero’s typical experiments involve engineering millions of protein sequence variants; recording their activity with high-throughput screening tools; and developing computer algorithms to parse through the generated data.
A key part of these experiments is droplet-based microfluidic technology, which uses a chip with many tiny channels filled with even tinier water droplets. Researchers can load biomolecules, chemicals or single cells into each of these droplets and watch them perform biological reactions.
“This technology lets us screen and analyze millions of protein variants in one fell swoop,” Romero explains. “The more data we collect and analyze with our algorithms, the higher the resolution at which we can understand the sequence-to-function mapping.”
Understanding that mapping has been Romero’s research goal since his PhD training in biochemistry at the California Institute of Technology. After completing his degree in 2012, he was a postdoctoral fellow in bioengineering at the University of California, San Francisco, and an assistant professor of chemical engineering at UCLA before joining the UW-Madison faculty in July 2016.
To develop his customized computer algorithms, Romero uses ideas from machine learning and artificial intelligence, which is why he continues to collaborate closely with chemical engineers, statisticians and computer scientists, in addition to fellow biochemists.
His lab is also home to a large and diverse group of trainees: Mark Politz, a postdoctoral researcher and UW-Madison graduate who trained with chemical engineer Brian Pfleger; two other postdocs (one with a PhD in chemical engineering, the other in chemistry); two chemical engineering graduate students jointly supervised with Pfleger; two biochemistry, one biophysics, and one cell and molecular biology graduate student; and several undergraduates.
With Politz, Romero has recently begun to study metabolic pathways—the next scale in biological organization after single proteins. Here, the aim is to understand and manipulate a series of chemical reactions that are controlled by the teamwork of multiple proteins. Since DNA contains the blueprint for building proteins, studying these complex pathways requires reading and writing very long stretches of DNA—a challenge that DNA sequencing technology is just now beginning to tackle.
For his first metabolic engineering project in Romero’s lab, Politz is using this new sequencing technology to coax Escherichia coli, a common bacterium in the intestine of humans and other warm-blooded organisms, into producing levodopa.
Levodopa (L-DOPA) is a versatile amino acid that is found in adhesives and solar cells and can also serve as a precursor of opioids. Of greatest interest to humans is that our body can turn it into dopamine, the main drug for Parkinson’s disease. By manipulating multiple different enzymes, Politz hopes to make levodopa from glucose and other naturally occurring small molecules in E. coli.
According to Romero, harnessing engineered E. coli for this job has practical advantages—such as cheaper drugs for Parkinson’s disease—over current production methods. But beyond that, the project is another step toward his long-term goal of getting microbes to convert basic sugars into any material of interest.
“We’re able to generate increasingly complex molecules from simply feeding sugars to microbes,” Romero says. “Since there are many chemicals that we cannot produce by traditional synthetic methods, or whose manufacturing is expensive or detrimental to the environment, I think the potential of metabolic engineering for a wide range of green chemistry applications is almost unlimited.”