Computational Modeling of COPA99aa-Antibody Complex for Novel Glioblastoma Multiforme (GBM) Detection and Therapeutic Applications
Glioblastoma Multiforme (GBM) is an aggressive, late-stage form of brain cancer that originates in the brain or spinal cord, with an immense impact on the patients and families affected by it. Less than 5% of the people affected by it survive for more than 5 years—known practices as of now are divided into 2 fields: diagnosis and treatment. The diagnosis phase starts with imaging in the form of CT (Computed tomography), MRI (Magnetic Resonance Imaging), or PET (Positron Emission Tomography). The treatment phase is usually chemotherapy with Temozolomide (TMZ), a drug, along with Radiation therapy in cycles. My research with the circCOPA99aa protein will be used as a molecular diagnostic in the diagnosis phase and as an inhibitor therapy in the treatment phase. circCOPA99aa is a novel protein linked to the early stages of GBMs. This can be considered a biomarker that would aid in detecting this aggressive brain cancer by inhibiting a feedback loop involving the NONO-SFPQ protein complex in GBMs.
An antibody is a protein that our immune system uses to defend against specific antigens (e.g., bacteria, viruses) inside us. It binds to them, allowing our T-cells (immune system attacker cells) and other immune system cells to attack the antigen. Recently, the role of the circCOPA99aa protein in the regression of glioblastoma disease has been identified and can be used as a biomarker to detect glioblastoma. I hypothesize that the antibody interacts with the binding site of the circCOPA99aa protein and thus can be used in a detection process. Therefore, I have performed computational modeling of the circCOPA99aa-antibody system. The amino acid sequence obtained from UniProt (a freely accessible resource of protein sequence and functional information) was uploaded to AlphaFold3 (An AI system developed by Google DeepMind that predicts a protein’s 3D structure from its amino acid sequence to achieve accuracy competitive with experiment) to get and understand its 3D structure. Then, HDOCK (A web tool for integrating template-based modeling and ab initio free docking of protein-protein and protein-DNA/RNA interactions) was used to determine the potential binding sites for antibodies on the CircCOPA99aa protein. Based on the visual analysis and binding energy calculations, antibody 1A5F (antibody tag, a way to identify antibodies) was chosen as the most appropriate antibody. This research could be very influential to the scientific community as it aids in detecting gliomas in patients in highly early stages and acts as a way to make treatments more effective, as finding antibodies that would bind to this would help us control cell division in the tumor.
Image depicting the control the CircCOPA99aa protein has on the NONO-SFPQ feedback loop and cell division in cancerous cells. If the CircCOPA99aa protein is produced in abundance within a cell, the NONO-SFPQ loop is running at a higher pace. When cell division occurs too quickly, the chances of mutations occurring are higher. When this happens, malignant tumors like GBMs form very quickly.
What was a challenge that you faced?
One of the major challenges I encountered was the lack of experimental (in vitro) validation. While I was able to build strong computational models of the COPA99aa-antibody interactions and simulate molecular docking, I faced barriers in accessing laboratory facilities to validate the binding experimentally. This limited my ability to fully translate computational predictions into biological confirmation.
How did you handle this challenge?
To address this, I focused on enhancing the computational rigor of my project. I optimized my HDOCK docking parameters, incorporated multiple prediction layers (like AlphaFold structures and antibody modeling), and attempted to run preliminary molecular dynamics simulations in parallel. These steps helped strengthen the credibility of my computational findings, even in the absence of wet lab validation.
If you or someone you know can help Kushal Patil take this to the testing stage in a lab, contact him at ikushal1990@gmail.com
What materials and software did you use?
- UniProt: The world’s leading high-quality, comprehensive and freely accessible resource of protein sequence and functional information
- AlphaFold: An AI system developed by Google DeepMind that predicts a protein’s 3D structure from its amino acid sequence to achieve accuracy competitive with experiment
- P2Rank: Machine learning based tool for rapid and accurate prediction of ligand binding sites from protein structure
- HADDOCK: The High Ambiguity Driven biomolecular DOCKing software is an integrative platform for modelling biomolecular complexes and is one of the most popular software for protein-protein and protein-ligand docking
- HDOCK Server: A web tool for integrating template-based modeling and ab initio free docking of protein-protein and protein-DNA/RNA interactions
- CHIMERAX: ChimeraX is a next-generation molecular visualization program developed by UCSF. It is designed for interactive visualization and analysis of molecular structures and related data
- PRODIGY: Prodigy is a web server designed for predicting the binding affinity in biological complexes
Student at Lincoln Sudbury Regional High School in Sudbury, MA
Edge on Science Alumnus
ikushal1990@gmail.com
Do you have a child interested in the biomedical sciences?
Register now for our BIOmedical Academy — limited spots available!
