New Method to Map Allostery Could Be Boon for Drug Discovery

Researchers in Barcelona have used deep mutational scanning to globally map allostery of two of the most common protein interaction domains and found that the number of potential therapeutic targets on the surfaces of human proteins is much greater than previously thought. Their approach enables research labs around the world to find and exploit vulnerabilities in any protein, including those previously thought undruggable, and could be a game changer for drug discovery.

“Not only are these potential therapeutic sites abundant, there is evidence they can be manipulated in many different ways. Rather than simply switching them on or off, we could modulate their activity like a thermostat. From an engineering perspective, that’s striking gold because it gives us plenty of space to design ‘smart drugs’ that target the bad and spare the good,” says André Faure, postdoctoral researcher at the Centre for Genomic Regulation (CRG) and co-first author of the paper.

Proteins play a central role in all living organisms and carry out vital functions such as providing structure, speeding up reactions, acting as messengers or fighting disease. They are made of amino acids, folding into countless different shapes in three-dimensional space. The shape of a protein is crucial for its function, with just one mistake in an amino acid sequence resulting in potentially devastating consequences for human health. 

Allostery is one of the great unsolved mysteries of protein function. Allosteric effects occur when a molecule binds to the surface of a protein, which in turn causes changes at a distant site in the same protein, regulating its function by remote control. Many disease-causing mutations, including numerous cancer drivers, are pathological because of their allosteric effects.

Despite their fundamental importance, allosteric sites are incredibly difficult to find; a protein might shapeshift in the presence of an incoming molecule, revealing hidden pockets deep within its surface that are potentially allosteric but not identifiable using conventional structure determination alone. Though some of the most effective types of drugs available, many allosteric drugs were discovered by accident.

Using double deep PCA (ddPCA), the CRG team generated the first ever map of allosteric sites in an SH3 domain and a PDZ domain. The method works by changing the amino acids that make up a protein, resulting in thousands of different versions of the protein with just one or two differences in the sequence. The effects of the mutations are then tested all at the same time in living cells in the laboratory.

“Each cell is a tiny factory making a different version of the protein. In a single test tube, we have millions of different factories and so we can very rapidly test how well all the different versions of a protein work,” says ICREA Research Professor Ben Lehner, Coordinator of the Systems Biology program at CRG and co-author of the study. The data collected from the experiments is fed into algorithms, which generate comprehensive maps that pinpoint the location of allosteric sites on the surfaces of proteins.

One of the great advantages of the method is that it is an affordable technique accessible to any research lab around the world. “It massively simplifies the process needed to find allosteric sites, with the technique working at a level of accuracy better than several different more expensive and time-consuming lab methods,” says Júlia Domingo, co-author of the study. “Our hope is that other scientists use the technique to rapidly and comprehensively map the allosteric sites of human proteins one by one.”

The authors believe that, if scaled up, the method could one day result in advances that can precisely predict the properties of proteins from their amino acid sequences, ushing in a new era of predictive molecular biology and allowing much faster development of new medicine.

“While some tools can predict a protein’s structure by reading its sequence, our method goes one step further by telling us how a protein works. This is part of a bigger vision to make biology as engineerable as aeroplanes, bridges or computers. We have faced the same challenges for over 70 years, but it turns out they are more tractable than we previously thought. If we succeed it will open a new field with unprecedented possibilities,” says Dr. Lehner.

The findings were published recently in the journal Nature.