Skip to main content


In silico drug design gives first model for selective PERK inhibition


Dr. Cardozo, the National Science Foundation (NSF)-funded researcher, and his team at NYU Sackler Institute have developed the first pharmacophore model for selective PERK inhibition as well as 14 selective PERK inhibitors using in silico drug design approaches along with experimental validation. This development could lead to novel anti-cancer drug leads against a fundamental cell survival mechanism. The generation of a pharmacophore model for selective PERK inhibition is a crucial milestone both for the science of the discovery and for the implications of the technological platform used to discover potent PERK inhibitors. The unfolded protein response (UPR) is a critical survival mechanism for tumor cells. Chemotherapeutic drugs may induce cellular stress in cancer cells that the UPR helps overcome, suggesting that inhibition of the UPR in tumors would be a novel anti-cancer approach to target the very stress response in cancer cells that hinder existing anti-tumor treatments. Among the three most prominent UPR transducers, the protein kinase PERK has a broader range of cellular effects. Most importantly, its critical role in tumor progression has been demonstrated by the observation that compromising PERK function inhibits tumor growth in mice. Therefore, inhibiting the kinase activity of PERK would be an important and novel therapeutic intervention in cancer. To date, however, no specific small molecule inhibitor of PERK has been identified.

This project relied on homology modeling of a protein or areas of a protein not available in crystal structure. One of the biggest challenges today is using computational tools to predict interactions in cases where there is no crystal structure or the crystal structure is a low resolution. Being able to model proteins or other structures accurately can allow researchers to analyze protein-ligand and protein-protein interactions for which crystallographers have been unable to get a clear structure. Dr. Cardozo’s team generated a model of PERK (PKR-like ER- localized eIF2α kinase) for their analysis. Initially, two homology models of PERK catalytic domain from two crystal structures of eIF2 kinase GCN2 (PDB code: 1zy4 & 1zy5) 31 were created, followed by conformational sampling of the activation loop. This ensemble of multiple receptor structures was then used to dock many compounds against it using Virtual Ligand Screening (VLS) (a computational screening procedure similar in concept to High Throughput Screening (HTS) but the results are analyzed and ranked according to the software’s binding-score function). If the homology model were inaccurate, it would have adversely affected the results of the VLS docking. The high hit rate of the VLS results (43.75%) compared to the significantly lower averages for HTS hits is of interest to anyone involved in drug discovery. Drug discovery is a long and expensive process. If computational modeling/simulation tools can make the early hit-to-lead stages faster and more accurate, then it makes sense for the drug discovery community to invest research into improving and disseminating these tools.

The three structural determinants that are important in establishing selective PERK inhibition were identified and laid out in the pharmacophore model. The most potent PERK selective compound utilizes three specific kinase active site contacts that, when lost in chemically similar compounds, abrogate the inhibition. The pharmacophore model is incredibly useful because it provides a blueprint or formula for searching for PERK inhibitors. Should later stages of the drug discovery process come up against problems of affinity, toxicity or bioavailability, researchers can potentially go back to the computational model and screen for other compounds that make the necessary contact points with PERK but have higher affinity, lower toxicity or are more bioavailable. This iterative process means that the early stages of drug development are less risky as compounds can be designed, improved and modified throughout the early stages of drug discovery.

This project was highly interdisciplinary, involving knowledge of computational biology in general and homology modeling and VLS specifically. It also required extensive understanding of structural biology. The team had to learn and master the tools and principles of all of these fields and integrate them into her model. This would not be possible without an interdisciplinary program, like the COB program, which gives students the opportunity to bring together several disciplines to create solutions to common and uncommon biomedical problems. The PERK inhibitors Dr. Cardozo’s team discovered and the pharmacophore they used to do so can be further developed and improved for both this project and projects for which it makes sense to use this strategy. This is a significant achievement made possible through IGERT.

Address Goals

This research project marks a significant achievement in Discovery. Finding new and innovative drug targets is crucial for generating treatments for diseases where treatment options are limited, including several types of cancer. The challenge for researchers is to come up with ways to rationally design new treatment options to assist or replace current drug options. Increasing our understanding of the fundamental principles of protein-ligand interactions is a crucial component in developing rational strategies for drug discovery. This project clearly lays out a technology platform and structure-based computational inhibitor design approach for generating PERK inhibitors and can be used as a model for how to develop pharmacophores for other drug targets. This finding could accelerate the PERK targeted drug discovery, a novel therapeutic intervention in cancer.

This project is also an excellent example for Learning, as this project has required that Dr. Wang master many different fields and integrate them seamlessly. These skills and experiences will be transferred to the people she trains as well as to other students that train in Dr. Cardozo’s lab. This research has been published and is publicly available, and there is a patent pending, ensuring that the compounds remain attractive for further development.