A Step Closer to Precision Oncology: Computational, Biochemical, and Cell-Based Screening to Find Compounds that Stabilize p53

Abstract

Personalized medicine in cancer aims to tailor a treatment plan that takes into account the unique features of a patient's malignancy. One therapeutic target that has a chance to affect a large population of cancer patients is p53. p53 is a tumor suppressor that activates senescence or apoptosis in cells that have accumulated mutations that could lead to cancer. Half of all cancers have mutations in p53, which highlights the importance of its role in disease. A subset of these mutations have been shown to inhibit p53 function by destabilizing p53's core domain. This led to the hypothesis that a personalized drug for patients with this type of destabilized p53 mutation could lead to apoptosis in cancer cells. There has been a lot of evidence supporting this hypothesis. This evidence has inspired many researchers to screen for small molecules that stabilize p53 mutants and rescue function. However, the hits discovered in these screens (with one potential exception) have not been found to be adequate drug leads for several reasons. Many have turned out to rescue function, but not by directly binding p53. Others bind p53, but either lack sufficient binding affinity or cause nonspecific cell responses. All of these are likely to induce side effects if used as part of a cancer therapeutic. This leads to the question: Is there a better way to find a small molecule stabilizer for cancer-associated mutants of p53? Here, I present an alternative approach that focuses on finding a direct binder to p53's core domain in order to avoid off-target effects. Our initial step was a computational approach that uses the crystal structure of p53's core domain in order to virtually screen a set of small molecules for binding. I found a novel pocket on the protein structure that I predicted to be druggable, because the site readily forms pockets during simulations of the core domain. I performed a virtual screen using the DARC, a docking tool from the molecular modeling suite, Rosetta, and selected the 28 best ranked compounds for biochemical testing with purified p53 using two different cancer-associated, destabilizing mutations. Surprisingly, I found that 11 of the 28 compounds stabilized both mutants. Further testing was done in cancer cell lines showing that 7 compounds activated p53 transcription of p21 and PUMA, which are known targets of p53. Using the fluorescent antibody pAb 1620 that binds natively folded p53, we showed that 4 of the compounds lead to a much higher concentration of folded p53 in cells. The excitingly high hit rate was found from a modest sized initial virtual screen of only 64,000 molecules. This suggests that this novel pocket is prone to bind molecules in a manner that rescues structure and function, and should be as a starting point for a larger screen. Also, the compounds from the current screen are intriguing hits that will be further analyzed and optimized to develop new stabilizers of p53.