Computational Design of an Activatable Yeast Cytosine Deaminase for the Advancement of Chemotherapy
Writer: Daniel Greenfield
Date: Spring 2016
Citation: Greenfield, D. & Khare, S. (2016). Computational Design of an Activatable Yeast Cytosine Deaminase for the Advancement of Chemotherapy. Rutgers Research Review, 1(1).
My name is Daniel Greenfield and I am currently a sophomore in the Rutgers School of Arts and Sciences Honors Program. My major is Biomathematics, and my minor is Computer Science. Since early 2015, I have taken part in research with Dr. Sagar Khare in the BioMaPS Institute for Quantitative Biology doing protein modeling and simulation. I'm very interested in research at the junction of mathematics, biology, and computer science, and specifically in research that may one day have an impact on patient lives. In the future, I hope to earn a Ph.D degree in some interdisciplinary field, as of now Computational Biology, but I will let my interests dictate my research path. After earning a Ph.D. degree, I plan on working in industry.
What is unique about the Khare Lab is the need to apply computer programming and algorithm implementation to biology, which personally was something I knew very little about entering college. The Khare lab seeks to understand the structural determinants of enzymatic specificity and reactivity using a combination of computational protein design and experimental characterization. The main goal is to develop an understanding of specificity at protein-ligand interfaces in order to induce desired mutations. These applications can range from chemotherapy in proteins involved with drug targeting to thermostability in proteins used in industry. The application of research to medicine and combating diseases is something I am beginning to develop a passion for. For example, cancer is one of the main diseases that kills humans, and is something I have always been interested in learning more about in order to help cure. My first research project opened my eyes to the wide-ranging impact that research can have on not just cancer, but all diseases.
A project focused on in lab is an attempt to advance chemotherapy through targeted drug delivery. Specifically, attempting to develop an activatable yeast cytosine deaminase protein through incorporating specifically chosen mutations to the protein. The hypothesis was that if we "closed" the protein structure to shut off the function of the protein by adding a specific protease recognition site and a linker region between the protein and the recognition site, then only a specific protease can cleave the protein to make it function. The beginning of the project entailed incorporating our chosen amino acids into the native protein structure, followed by applying the Rosetta (a molecular modeling suite) Kinematic Loop closure algorithm to rebuild the protein with the desired mutations incorporated. Then, the various protein structures outputted by this algorithm went through a Rosetta Design protocol in which we only allowed some amino acids to be mutated in order to relieve clashes between atoms of the amino acids that came as a result of the mutations and chain rebuilding. We only allowed certain amino acid residues to be designable because protein structure determines function, and too many designed or mutated amino acids could potentially alter the function. Then, we checked to see if any clashes still existed in the top 10% of models through a protein visualization software called PyMol. After "scoring" the redesigned proteins via a Rosetta scoring function, we sorted them based on the proteins whose newly closed loop had the highest solvent accessible surface area. This is vital to the project because if the closed protein loop is not easily accessible to solvent then proteases would have difficulty cleaving the loop of the protein in order to make it function. We then picked the top 5% of designs to test in the wet lab, which is still ongoing.
Figure 1: A protein model after the loop closure and design algorithms.
This modified cytosine deaminase will be useful because it will target chemotherapy to a specific tumor rather than exposing non-cancerous cells to chemotherapy, thus making chemotherapy less harmful to the patient. Hypothetically, this can be done by placing our modified cytosine deaminase into a tumor site and then giving the patient 5-fluorouracil (5FU), which is a non-toxic drug. Then, when the 5FU reaches the tumor site with the protein in it, it will be chemically modified to become 5-fluorocytosine (5FC), which is toxic and would kill the tumor cells.
The most significant results of this project as of now are my computational results. These results consist of both the proteins with the lowest score based on the Rosetta scoring function, indicating that the proteins have minimal clashing, and the proteins that have the highest solvent accessible surface area (SASA). We found that there was a tradeoff between the length of the linker region and the overall protein score and SASA score. The best model in terms of SASA and overall score generated by the computational algorithms we put it through had a linker length of two, with the two amino acids of our choice being after the protease recognition site (between the protease recognition site and the coordinating zinc atom of the protein). We also saw that in models with more linkers, the SASA would be around that of the best model, but that the overall score was significantly worse. In the future, it would be excellent if the proteins generated by the entire protocol can be expressed experimentally, however if not then I will focus on redesigning the proteins. My idea now is to focus on redesigning proteins with the spacer length of two as they generated the best data in order to see if any of those will end up expressed. I think that this project is important because it reflects the importance of disease related research outside of a hospital or doctors office. In the future, I hope to be able to expand this research project to other diseases.