I'm a computational scientist interested in the mechanisms of disease and ageing, including alterations in the structures and dynamics of proteins, their physiological regulation, and ways physics-based simulations and generative-AI approaches can help design therapeutics to reverse pathological molecular degeneration.
Here is a one-paragraph summary of my recent works:
My most recent paper with Dr. Zachary Levine, my doctoral advisor, outlines a strategy to drug intrinsically disordered proteins (IDPs) using a conformation-capture mechanism. IDPs are notoriously challenging targets, yet they represent over a third of human proteins and are implicated in many chronic human diseases. Unlike well-folded proteins, IDPs exhibit diverse conformations due to having low free energy barriers between different conformations. Therefore, traditional drug design strategies relying on experimental structures of proteins cannot be applied to drug IDPs. In this work, I propose an approach that takes advantage of the conformational flexibility of IDPs. We hypothesized that we could trap an IDP in a specific conformation if we had a molecule that could bind strongly enough to that conformation. I tested different compounds' affinities for a diabetes-associated protein called islet amyloid polypeptide (IAPP) fixed in the target conformation and a non-target conformation as a control. I tested several inhibitors (peptides, small molecules) previously known to bind IAPP and control molecules. Out of all the molecules, a class of oligoquinoline molecules, called 'foldamers,' were most effective at binding the target conformation of interest. Strikingly, this class of molecule was also found to be the most effective inhibitor at preventing IAPP's diabetes-associated aggregation, membrane-poration, and cell toxicity. This work demonstrates a proof of concept that IDP inhibitors can be designed around induced IDP conformations, stabilizing states not necessarily sampled natively. Crucial future work includes expanding this approach to larger IDPs and intrinsically disordered regions with biological regulatory functions (i.e., transcriptions). One assumption in this work was that all amino acids are equally accessible. This is mainly true for small peptides like IAPP but not necessarily for longer IDPs, which have more significant intra-molecular interactions. Nevertheless, many larger proteins contain regions of intrinsic disorder that are prime targets for these conformation-capturing molecules. For more information, see: https://doi.org/10.1101/2024.07.03.601611
Before this, I worked on a project to study the regulation of two pancreatic peptides, islet amyloid polypeptide (IAPP) and calcitonin-gene-related peptide (CGRP), secreted from the pancreas, and how their co-assembly affects insulin release. IAPP is a peptide that forms insoluble aggregates in pancreatic B-cells of most diabetic patients. During this process, it forms toxic intermediates primarily uncharacterized but thought to be composed of IAPP oligomers or co-assemblies of IAPP and other proteins. One candidate is the CGRP peptide, which is also expressed in the pancreas with high sequence homology to IAPP and competes for binding sites in the brain. Additionally, both peptides initiate pro-inflammatory actions, like dilating blood vessels. Yet the synergistic effects of IAPP and CGRP on insulin secretion and pathological aggregation in diabetes remained unclear. In this work, I simulated the conformational landscape of IAPP, CGRP, and the IAPP-CGRP co-assembly from replica-exchange molecular dynamics simulations. Most notably, IAPP was restricted to a subset of conformations while bound to CGRP, which was more stable than any conformation of IAPP alone. In this conformation, IAPP's aggregation 'hot-spot' was buried. These predictions were supported by our collaborators, who found that CGRP directly prevents the aggregation of IAPP fibrils. Interestingly, insulin secretion was further suppressed by IAPP-CGRP co-assemblies compared to IAPP or CGRP alone, revealing the peptide co-assemblies may be more toxic than the individual components alone. For more information, see: https://doi.org/10.1039/d1sc01167g
During my first year as a Ph.D. student, I worked on a project to understand the molecular basis for a genetic neurological disorder causing ataxia in mammals. This project was inspired by an accidental finding that a mutation in alpha-II spectrin causes a loss of muscle control in mice. Spectrin is a structural protein that maintains the shape and integrity of cell membranes. In certain neurological disorders, such as Alzheimer's, spectrin break-down products are often detected. Given that the spectrin mutation was located near a cleavage site, we hypothesized the spectrin mutation was somehow making the cleavage site more accessible, leading to larger quantities of break-down products. We suspected it may be due to spectrin's interaction with Calmodulin, which is bound near the cleavage site. To test this hypothesis, I made homology models of the mutated spectrin and the wildtype control and measured the binding affinities associated with Calmodulin binding. The results supported our hypothesis that the mutation decreases the affinity of calmodulin to spectin, resulting in spectrin's cleavage site being more exposed to proteases that can generate the breakdown products observed in neurodegenerative diseases. For more information, see: https://doi.org/10.1038/s41598-021-86470-1
My undergraduate thesis focused on simulating the extension of bacterial filaments called pili. Pili filaments are remarkable biomaterials that can stretch up to three times their length and pull forces as large as the bacterial body weight. However, the molecular properties that enabled them to stretch so elasticly were unclear. I used all-atom and coarse-grained simulations to study the extension of the pili, which was challenging to observe experimentally because the stretched state was hard to isolate. I characterized the stress-strain profiles of the filaments and found the first ever predicted Young's modulus for the Neisseria Menigitidus and Neisseria Ghonoreae type IV pili filaments. My simulations revealed their elasticity was similar to spider silk while being as strong as Nylon--a remarkable biofilament. One application of my work includes informing new antibiotics targeting extended pili filaments, which expose unique surfaces targetable by small molecules. Another is that the architecture of these filaments may be helpful for other bio-inspired materials, such as bio-nanotubes for drug delivery—hollowed filaments designed to deliver therapeutic cargo.
Other minor project contributions include:
Science Joke of the Month: Q. Why does the hamburger yield lower energy than the steak? A. Because it is in the ground state.
Style: Style and format inspired by James Guevara's website. Feel free to reuse this format, just leave a link back to my website!