All biological function depends on how different proteins interact with one another. Protein-protein interactions facilitate all the things from transcribing DNA and controlling cell division to higher-level functions in complex organisms.
Much stays unclear, nevertheless, about how these functions are orchestrated on the molecular level, and the way proteins interact with one another — either with other proteins or with copies of themselves.
Recent findings have revealed that small protein fragments have a whole lot of functional potential. Regardless that they’re incomplete pieces, short stretches of amino acids can still bind to interfaces of a goal protein, recapitulating native interactions. Through this process, they’ll alter that protein’s function or disrupt its interactions with other proteins.
Protein fragments could due to this fact empower each basic research on protein interactions and cellular processes, and will potentially have therapeutic applications.
Recently published in Proceedings of the National Academy of Sciences, a brand new method developed within the Department of Biology builds on existing artificial intelligence models to computationally predict protein fragments that may bind to and inhibit full-length proteins in E. coli. Theoretically, this tool may lead to genetically encodable inhibitors against any protein.
The work was done within the lab of associate professor of biology and Howard Hughes Medical Institute investigator Gene-Wei Li in collaboration with the lab of Jay A. Stein (1968) Professor of Biology, professor of biological engineering, and department head Amy Keating.
Leveraging machine learning
This system, called FragFold, leverages AlphaFold, an AI model that has led to phenomenal advancements in biology in recent times resulting from its ability to predict protein folding and protein interactions.
The goal of the project was to predict fragment inhibitors, which is a novel application of AlphaFold. The researchers on this project confirmed experimentally that greater than half of FragFold’s predictions for binding or inhibition were accurate, even when researchers had no previous structural data on the mechanisms of those interactions.
“Our results suggest that this can be a generalizable approach to search out binding modes which can be prone to inhibit protein function, including for novel protein targets, and you need to use these predictions as a place to begin for further experiments,” says co-first and corresponding creator Andrew Savinov, a postdoc within the Li Lab. “We are able to really apply this to proteins without known functions, without known interactions, without even known structures, and we will put some credence in these models we’re developing.”
One example is FtsZ, a protein that is vital for cell division. It’s well-studied but incorporates a region that’s intrinsically disordered and, due to this fact, especially difficult to review. Disordered proteins are dynamic, and their functional interactions are very likely fleeting — occurring so briefly that current structural biology tools can’t capture a single structure or interaction.
The researchers leveraged FragFold to explore the activity of fragments of FtsZ, including fragments of the intrinsically disordered region, to discover several latest binding interactions with various proteins. This leap in understanding confirms and expands upon previous experiments measuring FtsZ’s biological activity.
This progress is important partly since it was made without solving the disordered region’s structure, and since it exhibits the potential power of FragFold.
“That is one example of how AlphaFold is fundamentally changing how we will study molecular and cell biology,” Keating says. “Creative applications of AI methods, similar to our work on FragFold, open up unexpected capabilities and latest research directions.”
Inhibition, and beyond
The researchers completed these predictions by computationally fragmenting each protein after which modeling how those fragments would bind to interaction partners they thought were relevant.
They compared the maps of predicted binding across your complete sequence to the results of those self same fragments in living cells, determined using high-throughput experimental measurements wherein hundreds of thousands of cells each produce one sort of protein fragment.
AlphaFold uses co-evolutionary information to predict folding, and typically evaluates the evolutionary history of proteins using something called multiple sequence alignments for each single prediction run. The MSAs are critical, but are a bottleneck for large-scale predictions — they’ll take a prohibitive period of time and computational power.
For FragFold, the researchers as a substitute pre-calculated the MSA for a full-length protein once, and used that result to guide the predictions for every fragment of that full-length protein.
Savinov, along with Keating Lab alumnus Sebastian Swanson PhD ’23, predicted inhibitory fragments of a various set of proteins along with FtsZ. Among the many interactions they explored was a fancy between lipopolysaccharide transport proteins LptF and LptG. A protein fragment of LptG inhibited this interaction, presumably disrupting the delivery of lipopolysaccharide, which is an important component of the E. coli outer cell membrane essential for cellular fitness.
“The massive surprise was that we will predict binding with such high accuracy and, the truth is, often predict binding that corresponds to inhibition,” Savinov says. “For each protein we’ve checked out, we’ve been in a position to find inhibitors.”
The researchers initially focused on protein fragments as inhibitors because whether a fraction could block an important function in cells is a comparatively easy final result to measure systematically. Looking forward, Savinov can also be all in favour of exploring fragment function outside inhibition, similar to fragments that may stabilize the protein they bind to, enhance or alter its function, or trigger protein degradation.
Design, in principle
This research is a place to begin for developing a systemic understanding of cellular design principles, and what elements deep-learning models could also be drawing on to make accurate predictions.
“There’s a broader, further-reaching goal that we’re constructing towards,” Savinov says. “Now that we will predict them, can we use the info we’ve from predictions and experiments to drag out the salient features to work out what AlphaFold has actually learned about what makes inhibitor?”
Savinov and collaborators also delved further into how protein fragments bind, exploring other protein interactions and mutating specific residues to see how those interactions change how the fragment interacts with its goal.
Experimentally examining the behavior of 1000’s of mutated fragments inside cells, an approach often called deep mutational scanning, revealed key amino acids which can be accountable for inhibition. In some cases, the mutated fragments were even stronger inhibitors than their natural, full-length sequences.
“Unlike previous methods, we aren’t limited to identifying fragments in experimental structural data,” says Swanson. “The core strength of this work is the interplay between high-throughput experimental inhibition data and the expected structural models: the experimental data guides us towards the fragments which can be particularly interesting, while the structural models predicted by FragFold provide a selected, testable hypothesis for the way the fragments function on a molecular level.”
Savinov is happy concerning the way forward for this approach and its myriad applications.
“By creating compact, genetically encodable binders, FragFold opens a wide selection of possibilities to control protein function,” Li agrees. “We are able to imagine delivering functionalized fragments that may modify native proteins, change their subcellular localization, and even reprogram them to create latest tools for studying cell biology and treating diseases.”
