Share:

Using AI to Solve the Misfolded Protein Riddle

Using AI to Solve the Misfolded Protein Riddle

One of the most daunting challenges in drug discovery is finding drugs for proteins that misfold. The primary drug design objective is for a small drug molecules to be able to enter an active site and block an enzyme’s or ion channel’s normal function, inhibiting its action and causing the desired therapeutic response. The challenge of misfolded proteins is that they do not have traditional active sites; they are constantly changing confirmation, rendering conventional therapeutics useless. Alzheimer’s is one of the most prevalent of over 160 protein misfolding diseases, including other serious conditions such as neurodegenerative diseases, cancer, and diabetes.

In 2008, the Toronto-based Treventis Corporation began work on developing an Artificial Intelligence (AI) tool which can be used to aid in the therapeutic targeting of misfolded proteins. Over the next ten years, Treventis created a novel and sophisticated AI tool named Common Conformational Morphology (CCM), which screens a vast array of virtual compounds and produces 3D models of misfolded proteins. “CCM allows for accurate predictions of the confirmational morphology of intrinsically misfolded proteins,” says Christopher J. Barden, CEO and co-founder of Treventis. “The AI behind CCM focuses on the commonality of shape associated with misfolding.”

Protein production is a highly regulated cellular process that each cell monitors and surveys with an almost flawless precision. The most significant issues with protein misfolding are when the degradation process of misfolded proteins is disrupted. This disruption allows the misfolded protein to either aggregate, like the fibrils that form during Alzheimer’s, or becomes resistant to proteolysis, as seen in prion diseases. “What we know through structural experiments, at best, is something about what the fibrillar (completely misfolded) end-stage proteins look like in Alzheimer’s patients.” says Barden. “But much of the progression of Alzheimer’s happens due to the smaller oligomeric species. Knowing what the end-stage protein species look like, doesn’t help us to find a drug to bind to and inhibit the population of misfolded protein most responsible for disease activity.” Using the conventional methods of X-ray crystallography to find a structure-activity relationship (SAR) between small molecules and the protein’s active site has significantly advanced drug design; however, there are limitations. The most significant limitations of X-ray crystallography are its application in high throughput screening of large chemical libraries and the prediction of SAR. This is especially difficult when the protein undergoes numerous conformers making an X-ray near impossible. “Even with cryo-EM, we cannot count on seeing the pathogenic epitope of interest long enough in order to freeze it in place so that it can be visualized,” says Barden. “CCM starts with what was known about compounds that affected misfolding of more than one protein. Our ‘in silico’ investigations produced an engine for creating potential structures of small misfolded epitopes that could be exploited to do drug discovery.”

To date, CCM has been used to virtually screen over 1 million compounds filtered for blood brain barrier (BBB) penetrability. Augmenting the “software” is additional proprietary “wetware” in the laboratory, following up thousands of CCM hit compounds through cell-free experiments; thus, culminating in over 150 compounds with demonstrated active in vivo properties. “Our hit rate in our flagship tau program shows that approximately 2 out of 3 (66%) compounds identified/designed using our approach are active to stop misfolding in cell-free systems. That is a rate well above chance, which speaks well to the promise of CCM against targets like these.”

CCM has already been used by Treventis, alone and in collaboration with partners, to find putative binding sites and useful chemical starting points against a variety of different misfolding proteins. “We are always looking for new targets to which we can apply CCM,” says Barden.

Visit us at Axxiem.com

Canan Schuman, PharmD/PhD
Author: Canan Schuman, PharmD/PhD

Canan Schumann is Chief Editor for Axxiem and for Axxiem's blog "BiotechOntheWeb". When not writing for Axxiem, Canan works as a Clinical Research Scientist II at the Research and Development Department at Molecular Testing Labs, developing endpoint assays for the detection of infectious disease and cancer. Canan currently resides in Portland, Oregon, where he received his Honors Bachelor of Science (HBS), Doctor of Pharmacy (PharmD.), and his Doctor of Philosophy (Ph.D.) in Biopharmaceutics at Oregon State University.